Contribution
The book fixes an object for linguistics and a set of principles for it. Before it, linguistics was comparative and historical grammar with no account of what a language is at a moment. After it, there is a defined object (langue, a social system of signs, set apart from individual speech), a definition of the sign as a two-sided psychic unit with an arbitrary bond, a methodological split between the study of states (synchrony) and of changes (diachrony), and an account of meaning in which what a term is worth comes from its place among the other terms rather than from what it names. Linguistics is placed inside a projected general science of signs, sémiologie.
Key insight
A language is a system of pure values. Neither the sounds nor the ideas exist as units before the language divides them; the language cuts both “amorphous masses” at once, like a sheet of paper whose recto (thought) cannot be cut without cutting its verso (sound). So a term is not a sound attached to a ready-made concept. It is whatever the other terms of the system leave it: French mouton can share a signification with English sheep without sharing its value, because English also has mutton. Because the bond between sound and idea is arbitrary, nothing but these differences can hold a term in place, and nothing outside the system anchors it, which is also why the system can neither be reformed at will nor protected from drift.
Assumptions
- The object is constructed by the point of view (Intro. ch. III §1): “c’est le point de vue qui crée l’objet”. Langue is chosen as the norm for all other manifestations of language because it alone admits an autonomous definition; the argument is methodological, not empirical.
- Both sides of the sign are psychic (Intro. ch. III §2; Part One ch. I §1): the acoustic image is “l’empreinte psychique” of a sound, not the sound. Langue is located in the receptive and coordinating part of the speech circuit in the brains of a community; execution (phonation) belongs to parole.
- Langue exists completely only in the mass: a sum of imprints in each brain, “1 + 1 + 1 + 1… = 1 (modèle collectif)”, against parole’s “(1 + 1′ + 1″ + 1‴…)” (Intro. ch. IV).
- Spoken, not written, language is the object (Intro. ch. VI): writing is a second sign system whose only reason is to represent the first.
- A language state is an idealization (Part Two ch. I): a span in which change is negligible, compared to neglecting infinitesimals; “la notion d’état de langue ne peut être qu’approximative”.
- The argument works on words as stand-ins for units (Part Two ch. IV §1), although words are said not to coincide with the concrete units.
Key results
Principles, with where they are argued.
- The sign (Part One ch. I §1). The linguistic sign unites “non une chose et un nom, mais un concept et une image acoustique”; renamed signifié and signifiant, with signe for the whole.
- First principle, arbitrariness (§2). The bond is unmotivated with respect to the signified (“immotivé”), not free for the speaker. Onomatopoeia and interjections are set aside as few, partly conventional (French ouaoua, German wauwau) and drawn into ordinary sound change. Fully arbitrary systems “réalisent mieux que les autres l’idéal du procédé sémiologique”.
- Second principle, linearity (§3). The signifier unfolds in time in one dimension; “tout le mécanisme de la langue en dépend”. Accent is argued not to be a simultaneous second element.
- Immutability (ch. II §1). The signifier is “imposé” on the community, “la carte forcée”. Four reasons: arbitrariness leaves no rational norm to argue for a change; the multitude of signs; the complexity of the system; the inertia of a mass that uses language at every moment. “C’est parce que le signe est arbitraire qu’il ne connaît d’autre loi que celle de la tradition, et c’est parce qu’il se fonde sur la tradition qu’il peut être arbitraire.”
- Mutability (ch. II §2). Time also alters signs, always as “un déplacement du rapport entre le signifié et le signifiant” (Latin necare “kill” > French noyer “drown”). A language “est radicalement impuissante à se défendre” against such shifts, again because of arbitrariness. Even an artificial language (Esperanto) would escape its maker once in circulation.
- Synchrony and diachrony (ch. III). Sciences of value, like economics, must separate the axis of simultaneities from the axis of successions. Change strikes isolated elements, never the system as such (umlaut in gasti > Gäste, fōti > fēt), with no aim at what the new state will signify; the state that results is “fortuit”. The chess analogy: a position’s values depend on the pieces’ positions, one move shifts one piece, and a position is “affranchie de ses antécédents”; the analogy fails only in that the player intends the move and language does not.
- Synchronic and diachronic “laws” (ch. III §6). A synchronic law is general but not imperative (it states an arrangement, like trees in a quincunx); a diachronic event is imperative but not general. Sound laws are regular, but each is one particular event affecting a phoneme, not the words. Neither kind is a law in the juridical sense, and there is no panchronic viewpoint on particular facts (ch. III §7).
- Units are not given (Part Two ch. II–III). The spoken chain is a continuous ribbon; units appear only when sound and sense are delimited together (sižlaprã: “si je la prends” or “si je l’apprends”). Identity is that of the 8:45 Geneva–Paris express or a rebuilt street, not of a stolen coat; in such systems identity and value coincide (the knight replaced by any object given the same value).
- Value (Part Two ch. IV). Every value needs something dissimilar it can be exchanged for and similar things it can be compared with (a five-franc coin: bread, and other coins). Signification is the vertical link from image to concept inside the sign; value is the horizontal relation among signs, and the concept is “purement différentiel”, defined “négativement par ses rapports avec les autres termes”; “Leur plus exacte caractéristique est d’être ce que les autres ne sont pas.” Examples: mouton/sheep–mutton; redouter, craindre, avoir peur; the Sanskrit dual changes the value of the plural; louer against mieten/vermieten.
- The signifier is differential too (§3). What counts is the phonic differences; phonemes are “des entités oppositives, relatives et négatives”, shown by the free variation of French r against the fixed contrast of German r and ch. “Arbitraire et différentiel sont deux qualités corrélatives.”
- Only differences, but a positive sign (§4). “Dans la langue il n’y a que des différences sans termes positifs”, but this holds only of signified and signifier taken separately; their combination in a sign is “un fait positif”, and between signs there is not difference but opposition. “La langue est une forme et non une substance.”
- Two kinds of relation (Part Two ch. V–VI). Syntagmatic relations hold in praesentia along the linear chain; associative relations hold in absentia in memory (enseignement with enseigner, armement, éducation, justement). Choosing marchons! means choosing it against marche!, marchez! and against montons!, mangeons!; “l’idée appelle, non une forme, mais tout un système latent”.
- Absolute and relative arbitrariness (ch. VI §3). vingt is unmotivated, dix-neuf relatively motivated; motivation is never absolute. Languages range between “lexicological” (most unmotivated; Chinese as the extreme) and “grammatical” (Indo-European, Sanskrit); French is more arbitrary than Latin (ennemi against inimīcus).
- Grammar (ch. VII–VIII). Morphology, syntax and lexicology are not natural divisions; grammar should be ordered by the syntagmatic and associative axes. Abstract entities (the genitive, word order) rest on concrete units.
- Analogy is creation, not change (Part Three ch. IV, skimmed): honor is formed by proportion (ōrātōrem : ōrātor = honōrem : x) and coexists with honōs before replacing it; it counteracts the disorganising effect of sound change.
Claims
The support for nearly all of these is argument and illustrative example, not systematic evidence; the book is a reconstruction by its editors from students’ notes.
| id | claim | strength | support |
|---|---|---|---|
| C1 | Linguistics must take langue, the social system, as its object, separate from parole | moderate as a methodological choice; argued, not shown | Intro. ch. III–IV |
| C2 | The bond between signifier and signified is arbitrary (unmotivated) | moderate: cross-linguistic variation and the marginality of onomatopoeia; the argument is by example | Part One ch. I §2 |
| C3 | The signifier is linear, and the mechanism of language depends on it | weak to moderate: asserted as evident; the dependence is shown only through syntagmatic relations later | Part One ch. I §3; Part Two ch. V |
| C4 | Arbitrariness both protects the sign from deliberate change and leaves it defenceless against undirected change | moderate: a structural argument with historical examples | Part One ch. II |
| C5 | Synchronic facts and diachronic events are of different kinds and cannot be studied as one | moderate: argued through examples (umlaut, accent, Czech zero genitive) and the chess analogy | Part One ch. III |
| C6 | The value of a term is fixed by its relations to coexisting terms, not by its signification alone | moderate: cross-language and within-language examples; no counter-case examined | Part Two ch. IV §2 |
| C7 | In langue there are only differences without positive terms, yet the sign taken whole is positive | moderate, as a conceptual analysis | Part Two ch. IV §4 |
| C8 | Phonemes are oppositive, relative and negative entities | moderate: free variation where no contrast is at stake | Part Two ch. IV §3 |
| C9 | Every unit is held in syntagmatic and associative relations, and choice in speech is selection against a latent system | moderate: introspective examples | Part Two ch. V–VI |
| C10 | Relative motivation limits arbitrariness, and languages vary in its proportion | weak to moderate: illustrations, no measure | Part Two ch. VI §3 |
Concepts
- langage: the whole heterogeneous faculty and its exercise, physical, physiological, psychic, individual and social.
- langue: the social product of that faculty and the set of conventions adopted by the body of speakers; “le langage moins la parole”; a system of signs in which only the union of sense and acoustic image is essential.
- parole: the individual act of will and intelligence that uses the code, with the psycho-physical mechanism that executes it.
- signe, signifié, signifiant: the whole, the concept, and the acoustic image (a psychic imprint, not the sound).
- arbitraire: unmotivated with respect to the signified; not chosen by the speaker. Arbitraire absolu against arbitraire relatif (relative motivation through analysable composition).
- sémiologie: a science, not yet existing, of “la vie des signes au sein de la vie sociale”, of which linguistics would be a part.
- synchronie / diachronie: a state of the language and a phase of evolution; “idiosynchronique” is offered as the more exact term.
- valeur: what a term is worth by its relations to the other terms of the system, as against signification, the counterpart of the acoustic image inside one sign.
- rapports syntagmatiques / associatifs: relations in praesentia along the chain, and in absentia in memory; the associative ones are of indefinite number and no fixed order, except in inflectional paradigms, where the number is fixed.
- unité: a stretch of sound that is, to the exclusion of what precedes and follows, the signifier of a certain concept.
Connections
The book places itself against comparative grammar from Bopp to Schleicher, which it faults for never asking what its object was, and credits the Neogrammarians with treating language as a product of the collective mind. It takes from Whitney that language is an institution and its signs conventional, and goes further: arbitrariness separates language from every other institution, which Whitney did not see. It defends traditional grammar, Port-Royal included, as correctly synchronic. The economic analogy (value as exchange against the dissimilar and comparison with the similar) is explicit.
Inside this record the nearest readings are the compositional distributional models of meaning (LIT-273, LIT-272), in which a word’s meaning is fixed by its co-occurrence with other words. Saussure’s associative and syntagmatic relations, and value as position in a system, are a precursor of that idea, but the line from one to the other runs through later structuralist and distributionalist work this record does not hold. The connection is mine, not the book’s.
Bearing on the record
- THEORY-159 (filed from this reading). The record held no account of linguistic meaning as differential value. The thesis is filed as a Proposed THEORY with the book’s own limits stated: it is argued from examples, and the book itself makes the sign as a whole positive.
- LIT-273 and LIT-272. See Connections. No THEORY in the record rests on distributional semantics, so nothing is supported or contradicted.
- The LIT as filed. The LIT’s standard description says the value of a term is “purely differential”. The book confines this to signified and signifier taken separately and calls the sign as a whole a positive fact, with opposition, not difference, between signs. The LIT has been corrected.
- No instruction for machine-learning practice, so nothing for the anthology.
Limitations
- Not Saussure’s own text. The first-edition preface says Bally and Sechehaye, who attended few of the lectures, reconstructed the book as “une recréation” based on the third course and students’ notes, and arranged it “conforme à l’intention de l’auteur, même lorsque cette intention se devinait”. How far any formulation is Saussure’s cannot be told from this text; the published student notebooks and manuscript sources are other works.
- The edition read is not the 1916 one. See the history note.
- Argument by example. The principles are supported by chosen examples from Indo-European languages, chiefly French, German, Latin and Greek; no systematic test is offered, and counter-cases (sound symbolism beyond onomatopoeia, for instance) are dismissed as marginal.
- The tension between arbitrariness and the system is acknowledged but not resolved in the parts read: the system is where “une raison relative” reigns, yet it is also said to be immutable in itself.
- A linguistics of parole is promised and absent, as the editors say.
- Psychological claims (both sides of the sign are psychic; langue sits in the brain) are asserted, not investigated.
Open questions
- How much of the doctrine of value, and of the strong “only differences” wording, belongs to Saussure rather than to the editors? Reading the student notebooks or the manuscript sources would answer it.
- What would count against the claim that a term’s value is fixed by its oppositions? A case in which a term’s use is unchanged when a neighbouring term is added or lost would; the book gives the opposite cases (redouter) and none of this kind.
- Does the linearity principle survive for signed languages, whose signifiers are visual and can be simultaneous? The book contrasts acoustic signifiers with visual signals, which “peuvent offrir des complications simultanées sur plusieurs dimensions”, but does not consider a visual language.