Directions for Historical Linguistics: A Symposium

Winfred P. Lehmann and Yakov Malkiel, eds.

Saussure's Dichotomy between
Descriptive and Historical Linguistics

Winfred P. Lehmann

The University of Texas at Austin

The shape of linguistics today was set in large part by Ferdinand de Saussure. We agree with some of his principal insights into the make-up of language: its functioning as a semiotic system in which oppositions and interrelationships are of prime importance. On his views, whether these are original with him or carried on from previous scholars of language and sign systems, the generally observed subdivisions of linguistics are based: synchronic and diachronic, or as they are more generally labeled, descriptive and historical, as well as some of the offshoots of our discipline, such as sociolinguistics, which now are burgeoning. In his homage to Saussure, Meillet says that Saussure wished above all to mark the contrast between two manners of dealing with linguistic facts: the study of language at a given moment, and the study of linguistic development through time.¹ More matter of factly, Bloomfield states that Saussure “had for years expounded . . . the natural relation between descriptive and historical studies.”² Hjelmslev quotes with satisfaction the letter of Bally which applauded him for following the ideal formulated by Saussure.³ Whatever the deviations in further parts of their theory, they with other linguists followed Saussure in their approach to language as a sign system which can be studied from a synchronic or a diachronic point of view. It is the aim here to examine the dichotomy which Saussure propounded between synchronic and diachronic linguistics, to note the implications of it for him, and to observe some effects of the dichotomy on subsequent linguistic work until the present.

Classifying linguistics with other studies of valeur, like economics, which we refer to as the social sciences and humanities, Saussure ascribes to them two axes of study: one an axis of simultanéités, the other an axis of successivités.⁴ Because of the great number of signs in language and because of their diversity, linguistics above all other studies cannot investigate simultaneously the relationships in time and those in the contemporaneous system. It is therefore necessary to distinguish two linguistics (Cours, p. 116). After considering several possible terms — historical linguistics or evolutionary linguistics for the one and static linguistics for the other — Saussure gives his preference to the terms cited above, synchronic and diachronic.

Without doubt the dichotomy leads to oversimplification, even to awkwardness in the investigation of language. Saussure was far too capable a thinker to fail to see these; but he also recognized that in much scientific work progress requires simplification. In view of his impact on linguistics the simplifications he adopted, and bequeathed to succeeding linguists, must be specified, as must the difficulties themselves for which he proposed the simplification.

When we investigate any given language, such as German or French, the same material is handled twice, once by synchronic, once by diachronic linguistics. But how? The commonly adopted solution is to set up stages: Old French, Modern French, Old High German, Middle High German, New High German, and so on. For these stages Saussure provided an elegant name: espace de temps (Cours, p. 142). Admitting that one state of a language is not a point in time, he suggested that an espace de temps might be as long as ten years, a generation, an age, or more. The assumption of a “space in time” seemed tenable to him because of the manner in which languages undergo change: during some periods they may evolve little or undergo few changes, at any rate changes of little importance; during other periods they undergo a considerable number of changes rapidly. Periods with few changes would be handled as states; those of many changes would be selected for diachronic study, as the evolutionary phases of a language between its quiescent states. Even if there were a few changes in a period selected for synchronic study, these might be disregarded in somewhat the same way as mathematicians neglect infinitesimal quantities in some of their operations.

Obviously this identification of quiescent as opposed to evolutionary stages in language is artificial. It reminds us of a cyclical view of history, of Scherer's approach to literature with its alternation between stages of excellence and long periods of quiescence. Possibly it is unjustified to identify selected linguistic stages with periods of literary eminence. Yet the ideal langues which historical linguists have been using for their selected espaces de temps coincide remarkably with periods of literary glory. The high point of the Middle High German langue may be dated around 1200, the time at which the great medieval German poets were flourishing — with the apex of New High German and Old High German separated from it by approximately six hundred years. If we accept a cyclical view of human history, and if our literary standards agree with those of Scherer and his contemporaries, work in historical linguistics is greatly simplified. But our historical grammars then are regulated by nonlinguistic criteria. We produce a grammar for the Periclean Age of Greece, for the Augustan Age of Rome, for classical Sanskrit. Somewhat bluntly, historical linguists may center their activities around such periods because the langue at the time was indeed static — but for nonlinguistic reasons. A strong literary tradition had established a conventional language which was widely followed: Wolfram von Eschenbach would discard his East Franconian dialect divergences from the norm, as would an Alemannic Gottfried von Strassburg and a Bavarian Walther von der Vogelweide. The resulting langue gives the data for our Middle High German grammars; it is one of the espaces de temps in German.

In the welter of changing language we must obviously make some selection, especially when we need to get a tradition like the German language under control, and for pedagogical purposes; but we must also examine the basis for our simplification. Was the period from 1175 to 1225 really a period of linguistic calm in South Germany? Do we really find periods of linguistic calm coinciding with periods of literary excellence? Is our understanding of language in its development best obtained by selecting a number of stages which are defined by nonlinguistic criteria and using these as stages of reference?

The procedure of selecting espaces de temps came to be troubled when further investigation was made of language, notably by the dialect geographers. By concerning themselves with all types of language, not merely that of the eminent works of literature, they made it clear that even selected stages are not classically simple. The Middle High German around 1200 is not as neat as our elementary grammars may suggest. Nor is a precise distinction possible between various stages of language. Middle High German and New High German, Middle English and New English seem to merge at various times, depending on the criteria used to identify them — and similarly the stages of other languages.

Equally troublesome, forms of language contemporaneous with the ideal stages also share features with earlier and later stages. Language does not seem to evolve cyclically. Grimm's Law did not operate during one evolutionary period in proto-Germanic, to be quiescent for another period, and then resume its disruption of the German phonological system at another troubled period. Similarly, umlaut in the Germanic dialects, or the great English vowel shift, was not a momentary phenomenon, occurring between two static periods of langue. It is not necessary to pursue the details of these phenomena in the history of the Germanic dialects to establish this point. But it is important to observe that the linguistic data concerned are phonological. In distinguishing between various stages of a language, in pursuing diachronic linguistic study of any linguistic tradition, Saussure centered his attention around the phonological component of langue.

The discussion to this point has dealt largely with externals. Definition of stages of language by externals, by nonlinguistic cultural criteria, may be justified, for possibly we can grasp language only within a culture. If so, there is no great point in arguing about the basis of selection of any espace de temps used in historical linguistic study. Because literary and language study have been closely associated, we may take as our selected points in the history of a language those at which eminent literature was produced. Alternatively we might select periods of great religious interest, such as the Cluniac reform, the Protestant revolution; or of technological change, such as the improvement of communication systems through the Carolingian reform of the writing system, the introduction of the printing press, and so on. If language is indeed treated as a subordinate component of culture, and if our criteria for selecting espaces de temps had no intimate relationship with language itself, the classification we arrive at after agreeing on our criteria would not distort our views of it. But for Saussure the basis of selection of important diachronic stages was intimately related to his view of language. His distinction between synchronic and diachronic study was directly associated with his understanding of language. It is because of this intimate connection between his understanding of language and the two types of study, synchronic and diachronic, that a scrutiny of his dichotomy is of fundamental importance in linguistics.

For besides proposing two manners of study for dealing with linguistic facts, Saussure saw in the mechanism of language, in the signifiant, two strata, or levels: a phonological and a morphological.⁵ Whatever terms we adopt for the two levels, or whatever terms he himself used, the signifiant of language is made up of two types of components: one, the components of sound, phonemes; the other the components of form, morphemes. And each set of components makes up a system.

For Saussure each set can be studied variously. The set of sounds may be examined under phonologie — which has no relation to time; one may deal with the production of sound, the physical events, and the auditory effects. Sounds may also be studied from the point of view of their role for the native speakers of any language; of the three types of phonological study, this type most resembles subsequent phonemics. A third type of study, phonétique proper, deals with sounds as they change in time, belonging in contrast with the other two types to diachronic linguistics.

Grammatical statements on the other hand are synchronic. They may display the result of diachronic changes, as in Latin facio: conficio (Cours, p. 219); but it is even dangerous to use for such pairs the term permutation. For such a term may provide a “false notion of movement where only a state is present” (Cours, p. 219).

It is static (descriptive) linguistics, or the description of the state of a language, for which Saussure reserves the term grammar (Cours, pp. 185ff.). Grammar deals with langue as a system of means of expression; “grammatical” is equivalent to synchronic and significant . . . there is for us no “historical grammar.”⁶ Grammar, accordingly, the study of the morphology and syntax of a language, must be assigned to descriptive linguistics, excluded from diachronic linguistics.

The primary aim of diachronic linguistics is therefore phonétique. Phonétique does not deal with significance, nor with a grammar; in giving the history of the sounds of a word, one can disregard its meaning. And if the evolution of a langue is reduced to that of sounds, the opposition between the objects proper to the two divisions of linguistics becomes clear: diachronic is equivalent to nongrammatical as synchronic is to grammatical (Cours, p. 194). There is a precise dichotomy.

In concluding his general chapter on diachronic linguistics, however, Saussure states that phonétique may not account for all the developments in a language; after one has applied it, one finds a residue which seems to justify the notion of a “history of the grammar” (Cours, pp. 196-197).⁷ But the intricate explication which a detailed statement of the distinction between diachronic and synchronic would require is beyond the scope of the Cours. (The footnote of the editors at this point ascribing the problem to Saussure's lack of consideration of la linguistique de la parole probably attempts too simple an explanation. For Saussure's subsequent discussion embraces the possibility of competing forms in the parole before they are incorporated in the langue). The Cours itself maintains the distinction, examines the mechanism of interplay between phonétique and grammaire, and firmly establishes two linguistics: synchronic and diachronic.

We may clarify the distinction by examining the topics discussed in the third part of the Cours, for they follow from Saussure's view on the change of language. First, phonetic change takes place. This may have an effect on the grammar, disrupting grammatical ties. The ties are then restored, often rearranged, by the workings of analogy; analogy is at once a source of renovation and of conservation. Through analogy, then, the grammar of a language may be rearranged, but “analogical phenomena are not changes”; they correspond rather to creations in a synchronic stage of language. Even though analogy acts as an agent in modifying grammaire, the action belongs on the synchronic plane. The dichotomy between phonétique and grammaire, between diachronic and synchronic linguistics remains firm. It not only establishes two manners of dealing with linguistic facts, two types of linguistic study; it also divides language itself into two components.

Like most positions in a science which later become fixed as dogma with their own inadequacies, Saussure's dichotomy represented an advance over that of his predecessors. In a fine passage (Cours, pp. 118-119), Saussure objects to the linguistics of his time, which he ascribes largely to Bopp, as being entirely absorbed in diachrony. Moreover, it never defined its aims. Comparison was only a means of reconstructing the past. Its basis too was poorly defined. With comments particularly relevant today, Saussure contrasts nineteenth-century linguistics with that it ousted, traditional grammar, whose most notable work he recognizes as the grammar of Port-Royal. This Saussure esteems because it attempted to describe the state of French at a given time, the period of Louis XIV. Further, although it erred in neglecting parts of language, such as derivation, and in being normative rather than determining the facts, its basis is less subject to criticism and its aim better defined than are those of nineteenth-century linguistics. Accordingly there were few grounds for the reproach that it was not scientific. Saussure forecasts that linguistics will return to the static view of traditional grammar, after its rejuvenation by historical linguistics. Presumably it was to carry out this aim that he composed his course of lectures on general linguistics.

Unquestionably Saussure achieved his aim. But we have suggested that he did so at a cost: in the resulting linguistics, stages of language were artificially set up which were relatively static, with intervening periods of considerable change. Though pedagogically useful, this view of language also has its shortcomings. And we may now be at a point when, having profited from the contributions of historical linguistics, we can resume relations with eighteenth-century linguistics, and build on the new static linguistics a new historical linguistics.

The historical linguistics which Saussure proposed tremendously emphasized phonology. By it the essential change in langue is at the phonological level. To be sure, this change disrupts the grammatical system, leading to adjustments at the morphological level. But these adjustments are of a different order from phonological change. They are readjustments rather than innovations. Moreover, in Saussure's historical linguistics semantic change goes almost completely unmentioned. The signifiant of language makes up the central concern of Saussurean linguistics, and in historical linguistics the phonological component of the signifiant. It is scarcely surprising that subsequent work has moved to modify this view. But as yet there has been none of the explicit, general formulation which made Saussure so influential. The time may now be at hand for such a formulation, based on the notable work that has been carried out after the publication of the Cours.

To illustrate views which are departures from Saussure one may cite a segment from Szemerényi's statement on recent etymological research; Szemerényi sees in etymological work from about the beginning of our century a “break away from the pursuit of original forms and/or meanings, a shift of emphasis to historical processes” (p. 177).⁸ This break may partially accord with Saussure's views; but the shift to a concern with processes indicates that etymologists are no longer chiefly concerned with determining items or states, rather with the procedures by which interrelations in language are achieved. We may exemplify the shift of interest with one of Saussure's comparisons, that between an observer of language and an observer of a game of chess (Cours, pp. 126-127). Saussure points out, and we agree, that for anyone observing a game of chess at any given point it is unimportant to know how a given position of the pieces has been arrived at. For a description of the position at the moment, the history of previous moves is unessential. Similarly, if a linguist were an external observer of language, a synchronic description accompanied by a subsequent diachronic statement would be adequate. But the role of observer no longer satisfies linguists. For just as an observer of a chess game at any one point obtains only a partial view of the total game, so observation of selected points in the development of a language yields an inadequate account of language in its development. In historical as well as in descriptive linguistics an approach is now favored which deals with language from the point of view of the participant, to whom the moves are of central importance, whether in following the course of a game or in interpreting the situation at the moment.

It has been noted above that Saussure's view of a linguist as an observer leads to a concern for language as a state. Today there is a far greater concern for language in operation, as will be discussed below. An observation on some of the current work in linguistics, that dealing with linguistic theory, indicates briefly the shift of concern: “The current controversy in linguistics is largely concerned with the question of what is the best method of describing both states and operations.”⁹ Although Thomas was referring to work in descriptive linguistics, the quotation from Szemerényi illustrates that the concern extends to historical linguistics as well.

Interest in operations is evident at the morphological and semantic level, as well as at the phonological. But if we occupy ourselves primarily with operations, not states, language is highly fluid. To a great extent the procedures we apply are also less general; the term law for example differs from its use for physical sciences even more than it did for Grimm. And change may refer to minute shifts. Any new presentation dealing largely with operations, a new course of general linguistics, will be far less neat than is Saussure's, but it may also comprehend more details of linguistic events.

The shift to a more fluid phonology has been so prominent that it needs little elaboration. Jakobson proposed guidelines for a more flexible handling of phonology in the early years of the Prague Circle.¹⁰ His “principles of historical phonology” lead to a view of regarding successive mutations in language from either a synchronic or a diachronic point — and at any time, either before, during, or after the mutation, we may look at the language synchronically or diachronically. Already in his article of 1930 Jakobson introduces such terms as chain, which was exploited at length in possibly the most extensive discussion of the sound systems of language regarded from the point of view of the processes underlying their changes.¹¹

The greater fluidity in handling phonology may be the theoretical deviation from Saussure's approach which has been most widely discussed by historical linguistics, but it is also one of the least. Rather than adopt major stages for synchronic description, it suggests seizing any period of language. It may therefore be primarily substituting linguistic realism for pedagogical selectivity. But as in Saussure's diachrony, the major concern of this historical linguistics is with phonology as the innovating segment of language — maintaining in this way Saussure's link between diachronic linguistics and the phonological level of language. Since an assumption of change rather than rearrangement at other levels of language would be a far greater departure from Saussure, these will result in a far more different approach to language than is Jakobson's “integral method.”

The most explicit attempt to set up principles of a dynamic morphology in historical linguistics is that of J. Kuryłowicz.¹² In this article analogy is viewed as a dynamic mechanism: the morphological patterns of language seem to stand in some kind of tension, which may be resolved in accordance with six rules. Some of these rules predict that one of two or more parallel forms may replace another; they are general principles, virtually laws on potential change in morphology.

In contrasting the views of Kuryłowicz and other contemporaries¹³ with those of Saussure we may recall how explicit Saussure was about the ineffective role of the speaker in initiating, and even in controlling, change in language. His comparison between language and chess reflects this view, for he states that “the chess player has the intention of carrying out a shift and of effecting an action on the system; but language does not premeditate anything.”¹⁴ And at the end of the section he speaks of a blind force operating against the sign system. Kuryłowicz in contrast views the speaker as deciding between alternate forms, and through such decisions controlling the selection of two or more competing forms. We may characterize the aim of Kuryłowicz as an attempt to specify how the speaker controls the effects of change in language. Others, as do Weinreich, Labov, and Herzog in their contributions to this volume, are attempting to specify the mechanisms of the “blind force” that operates against the system of signs.

The study of operations, or processes in historical linguistics may then have enlarged on Saussure in dealing with problems of language change that he left untouched. A clearer specification of analogy and, even more, attempts at identification of the source of language change within the system modify considerably the position of Saussure. They also promise to illuminate questions left untouched by earlier historical linguists, and to enlarge our understanding of language in its change. If we contrast an excellent grammar of the past, such as W. Streitberg's Urgermanische Grammatik (Heidelberg: Carl Winter, 1896), we may note that it primarily describes states, and confronts a succeeding state with one that had preceded it. When for example we compare with Streitberg's treatment of the Germanic consonant shift that of E. Prokosch in A Comparative Germanic Grammar (Philadelphia: Linguistic Society of America, 1939), we note in it a much greater concern with processes. For Prokosch, however, the chief interest lay in extralinguistic processes, such as the social setting at the time of the shift and articulation of the sounds. Subsequent study, following in great part J. Fourquet in Les mutations consonantiques du germanique (Paris: Les Belles Lettres, 1948), has attempted to specify the linguistic processes involved in the Germanic consonant shift. It would require a great deal of space to give in detail the changing approach to the Germanic consonant shift and to other problems of historical linguistics; of prime concern here is that, as in the study of etymology, increasing emphasis has been placed on processes, in the attempt to account for change in language.

A comprehensive attempt to broadly examine language in change has been displayed in the work of Malkiel. Dealing in an impressive set of studies with language at all of its levels, he has introduced flexibility in the interpretation of change. His approach agrees to a large extent with that of Saussure, for some of the topics he discusses — weak sound change, trajectories — suggest basic concern with phonological change. Yet for Malkiel analogy may launch developments, the pressure of a morphological pattern may block or wipe out developments, grammatical as well as phonological structure may bring about change. As his essay in this volume illustrates, he is directly concerned with a dynamic effect of grammar, and with this concern he departs from a strict Saussurean approach.

A greater departure, however, will be introduced by linguists who relate all parts of language to syntax — who rigorously determine interrelationships between all linguistic elements and the unit they see as central to language, the sentence. The assumption that the sentence is the basic entity of language is not new; Saussure commented on it (Cours, pp. 148-149, 172-173), but rejected it on the grounds that the sentence belongs to parole, not langue.¹⁵ An approach to language ascribing such a role to the sentence would therefore bring about a considerable change from Saussurean linguistics. If the total emphasis is further placed on processes, not states, the departure from a Saussurean approach will be far greater than is that in the work of Malkiel, Kuryłowicz, and others. Since little work in diachronic linguistics has been undertaken with a transformationalist approach, however, the possible great distinction between some historical studies of the future and that basing itself on Saussure is merely indicated now. Yet until such works are published, there is little point in discussing theoretical problems which will confront a syntactically based phonology.

The usefulness of these approaches will vary from one instance to another. Possibly each approach will clarify specific problems in language, and, following Malkiel's study “Each Word Has a History of Its Own,” we may even suggest that fractions of various problems in one language may yield to each approach.

By a distinctive feature analysis, for example, we may specify more precisely the twofold development of PGmc au to OHG ou before labials and velars, to ō before dentals and PGmc h: An acute distinctive feature in the following consonant is accompanied by lowering of the second formant and fixing of both formants; the grave consonants regulate lowering of the first formant in the early part of the diphthong. Such an interpretation would require us to specify OHG r as acute, as well as PGmc h. This specification of the process underlying a sound change makes use of distinctive features on a syntagmatic plane. Other specifications, such as those of processes in the Germanic consonant shift, would primarily use distinctive features on a paradigmatic plane.

Still other procedures, at this time much more tentative, might deal with problems that have eluded analysis from a Saussurean dichotomy between phonology and morphology. It has been proposed, for example, that the problematic -ē of the Gothic genitive plural may be ascribed to analogy.¹⁶ For it is found in nouns with front vowel characterizing the genitive singular, for example, dags, gen. sg. dagis (gen. pl. dagē), while -ō is found with back vowels in the genitive singular, for example, giba, gen. sg. gibōs (gen. pl. gibō). This situation arose in a phonological system with four long vowels:

Speakers might then have regarded the two low long vowels as not distinctive in the environment of the genitive suffix and may have distributed them in accordance with the pattern of the vowels in the genitive singular.

Accordingly, besides expecting modifications in the strict Saussurean dichotomy, and solutions to problems which it left unsolved, we may now be at a point of dealing with areas of language that seemed unpromising to Saussure, such as the area of semantic change. For Saussure, changes in meaning were individual, and accordingly incapable of general treatment.

We may illustrate with his example, poutre (Cours, p. 132), which at some time was changed in meaning from ‘filly’ to ‘beam, support.’ Apparently this change cannot be related to other semantic changes, particularly if we regard the meaning of ‘filly’ as an indissoluble unit. Similarly, other examples, such as NE ‘horse,’ which is also used for supports, as in ‘carpenter's horse, clotheshorse.’ Yet although we may not provide generalizations for such changes in meaning, we find it difficult to believe that some entities of the language might be extended in meaning as was horse in these ways, for example, lungs or water or three or 's. Noting such examples, we may suggest that eventually general statements on change of meaning may be found in a more thorough analysis of units selected for study of meaning. If we determine for horse the distinctive features of meaning, and their place in the meaning structure of a given language, we may be able to specify semantic shifts in somewhat the same way as we do phonological shifts. Just as we specify a modification of p to f, or e to i, on the basis of a distinctive feature approach, applied to the structure of a specific language in its development, so we may account for a modification of ‘sturdy animate quadruped’ to ‘sturdy inanimate quadruped’ and the like in a given semantic system. To be sure, in accounting for changes in a phonological system, a small number of distinctive features are adequate, whereas in a semantic system thousands would be necessary. Yet this approach to locating general principles in a further area of language shift is not contrary to Saussure's views; rather, it is an extension of them.

When modifications in historical theory are proposed, their proponents might with advantage examine the views of Saussure. Such a course may be useful because Saussure may already have examined the proposed modification. Further, highly careful studies, such as Benveniste's on Saussure's views on the arbitrariness of the linguistic sign (Acta Linguistica 1 [1939]), have not overturned the position of Saussure, though they have refined it.¹⁷ Other more recent studies may represent somewhat further departures from Saussure's principles, particularly from his insistence on a strict dichotomy. But much of the activity in descriptive linguistics, whatever it is labeled, is merely leading to a finer analysis of language, and will in this way permit more precise definition of phenomena of concern to historical linguistics, finer accounting of change in language. Other areas of vigorous activity in linguistics are concerning themselves with relatively unstudied segments of language. Still other activity is closely examining language in its social setting. If like Saussure we may attempt a clarification of linguistic activities by comparing those in another science — without suggesting parallelism in techniques or subject matter — we may note a great amount of work parallel to that in molecular biology, on the one hand, and to ecology, on the other, in the linguistic examination of minute entities and the role of language for the learner, the speaker, and society. While this work, and the mode of specifying language, is bringing about modifications in linguistics, and accordingly departures from Saussure, it does not nullify his entire theory. The departures which have been introduced, particularly in reducing the distinctions between his dichotomies, may lead to elimination of shortcomings of his linguistic theory, in somewhat the same way as were those of Bopp and his successors by Saussure, without leading to positions which disregard the achievements of past theory.

Footnotes

¹ A. Meillet, Linguistique historique et linguistique générale, II (Paris: Klincksieck, 1938). His statement on the work of Ferdinand de Saussure is given on pages 174-183. The excerpt is on page 183: “l'étude de la langue à un moment donné, et l'étude du développement linguistique à travers le temps.”

² L. Bloomfield, Language, pp. 18-19 (New York: Holt, 1933).

³ L. Hjelmslev, Essais linguistiques, p. 31 (Copenhague: Nordisk Sprog- og Kulturforlag, 1959).

⁴ Ferdinand de Saussure, Cours de linguistique générale. Third edition, edited by Charles Bally and Albert Sechehaye with the collaboration of Albert Riedlinger, p. 115 (Paris: Payot, 1949). References will be to this text. Commentary on Saussure's views is voluminous, and apparently will continue. Although additional materials of his have been published, such as Les sources manuscrites du ‘Cours de linguistique générale’ de Saussure, by Robert Godel (Genève: Droz, and Paris: Minard, 1957), I am dealing only with the Cours and its influence on linguistic work of the time after Saussure. Other points of view of his, and his way to them, may be valuable in providing further insights into language, and into the perspectives which are essential for work of lasting importance in linguistics, but especially into his own intellectual development. For a statement on his fundamental views which is unlikely to be surpassed see É. Benveniste's essay, “Saussure après un demi-siècle,” now published as Chapter III of his Problèmes de linguistique, pp. 32-45 (Paris: Gallimard, 1966).

⁵ Saussure's views on the treatment of sounds in language are too complex for detailed discussion here. Moreover, they have been discussed in detail, as by R. Wells, “De Saussure's System of Linguistics,” Word 3.1-31 (1947); reprinted in M. Joos, Readings in Linguistics, pp. 1-18. It is almost ironic to read today Wells' unhappiness about Saussure's inconclusiveness concerning phonemics, though with Wells' other comments these illuminate widespread linguistic views of the nineteen-forties as well as Saussure's.

⁶ Cours, p. 185: “qui dit grammatical dit synchronique et significatif . . . il n'y a pas pour nous de ‘grammaire historique’.”

⁷ It may be this residue which is yielding to subsequent study that contrasts with the approach of Saussure, as in the work of J. Kuryłowicz and Y. Malkiel, among others.

⁸ O. Szemerényi, “Principles of Etymological Research in the Indo-European Languages,” Innsbrucker Beiträge zur Kulturwtssenschaft, Sonderheft 15, Innsbruck 1962. Sonderdruck aus II. Fachtagung für indogermanische und allgemeine Sprachwissenschaft, pp. 175-212.

⁹ O. Thomas, Transformational Grammar and the Teacher of English, p. 15 (New York: Holt, Rinehart and Winston, 1965).

¹⁰ See his “Principes de phonologie historique,” Selected Writings I. Phonological Studies, pp. 202-220 ('s-Gravenhage: Mouton and Co., 1962). The essay was first published in 1931, though presented in 1930.

¹¹ André Martinet, Économie des changements pbonétiques. Traité de phonologie diachronique (Berne: Francke, 1955).

¹² These principles are outlined in his article “La nature des procès dits ‘analogiques’,” Acta Linguistica 5 (1945-1949), 15-37. They have been elaborated in Chapter I of The Inflectional Categories of Indo-European (Heidelberg: Winter, 1964). See especially pp. 38 ff. Examples of their application by Professor Kuryłowicz may be found in this book, and others to which he refers in his discussion.

¹³ O. Szemerényi, Studies in the Indo-European System of Numerals (Heidelberg: Carl Winter, 1960).

¹⁴ Cours, p. 127. “Il n'y a qu'un point où la comparaison soit en défaut; le joueur d'échecs a l'intention d'opérer le déplacement et d'exercer une action sur le système; tandis que la langue ne prémédite rien ...”

¹⁵ See especially Cours, p. 172: “On pourrait faire ici une objection. La phrase est le type par excellence du syntagme. Mais elle appartient à la parole, non à la langue.” J. J. Katz and P. M. Postal, in An Integrated Theory of Linguistic Descriptions (Cambridge, Massachusetts: The M. I. T. Press, 1964), state on page ix that it is necessary to distinguish sharply between language and speech, referring to Saussure; but in dealing with “language” they take as their unit the sentence.

¹⁶ W. P. Lehmann, “The Gothic Genitive Plural -ē: Focus of Exercises in Theory,” Papers in Linguistics in Honor of Léon Dostert, edited by William M. Austin, pp. 108-111 (The Hague — Paris: Mouton, 1967).

¹⁷ “Nature du signe linguistique,” Chapter IV, pp. 49-55, of his Problèmes de linguistique générale (Paris: Gallimard, 1966).