Saussure’s sound symbolism

John Joseph
University of Edinburgh

“The most celebrated opponent of the sound symbolic hypothesis,” writes Magnus (2013: 201), “was, of course, Ferdinand de Saussure”. Of course. One of Saussure’s key contributions to modern linguistics is the principle of the arbitrariness of the link between sound and meaning, or more precisely between signifier and signified within the linguistic sign, his most detailed discussion of which took place in his third course in general linguistics in 1910-11. It was carried over into the posthumous Cours de linguistique générale (1916), where it has long been the target of attacks by linguists convinced of the explanatory power of sound symbolism.

But how is it then that in the last paper he published during his lifetime, Saussure (1912) argued that a group of Latin adjectives had developed in a particular way because the shape of the diphthong in their stressed syllable is mimetic of the shape of the idea common to the words containing the diphthong?

Few investigators of iconicity have read the 1912 paper (one of the rare later studies to discuss it is Gmür 1990: 47-49). Like Magnus, they generally understand sound symbolism to be the direct opposite of the arbitrariness which Saussure professed. Since he wrote this paper just after, or even while, giving the lectures on arbitrariness, it may look like a deathbed conversion. But given that it was for a Festschrift presented to Vilhelm Thomsen on his 70th birthday, 25 January 1912, chances are that Saussure was writing the paper no later than the semester in which the lectures on arbitrariness took place. If we go back 35 years to his second published paper (Saussure 1877), it too proposed a form-meaning link, more conventional but still of the sort that typically gets classified as iconic (see Joseph 2012: 200-202). In between, we have his testimony from 1892 concerning his own synaesthetic associations of vowels with colours, textures and smells (Joseph 2012: 392-397).

This is starting to look like quite a different man from the one of whose limited vision Magnus & Co. despair. In this post I shall explain why I think their despair is misplaced.

The 1877 paper concerns the diachronic development of the Indo-European languages, and does not directly address the question of speakers’ awareness. The same may be said of the 1912 paper, which opens:

The diphthongs ai and au occupy only an ill defined place within Indo-European morphology or vocabulary. Among other facts which contribute to their obscurity, they have figured only in a group of words extremely weak since the origin […]. Taken individually, these words in turn very often have an isolated position in the language, being attached neither to a strong verb nor to any etymological family whatever. It is clear that this latter feature, to the degree that it would confer upon these words a certain sort of regularity, does so only in a wholly exterior and negative way.

Saussure (1912: 202 [1922: 595])[1]

The ‘group of words’ in question are adjectives linked phonetically by having a diphthong that starts with /a/, and semantically by referring to some infirmity or deviation from the ‘right’ or ‘straight’. The diphthong could be /ai/ (as in Latin caecus ‘blind’) or /au/, but also /ar/, /al/, /an/ or /am/, all of which are analyzed by Saussure, starting in his Mémoire (1879), as /a/ + sonant, hence as diphthongs in the same way that /a/ + /i/ or /u/ form a diphthong.

The sound symbolism inheres in how the ‘straight’ vowel /a/ ‘deviates’ off into the sonant. Saussure points out that words such as Latin blaesus ‘stammering’, claudus ‘limping’, calvus ‘bald’, mancus ‘maimed’, are very few in number and are isolated within the language, being attached neither to any strong verb nor to an etymological family. Normally, this would be a reason for not studying them at all. But he maintains that their isolation and the rarity of the /a/ diphthongs they contain give them ‘a certain kind of regularity, though it does so only in a completely exterior and negative way’. The semantic link, on the other hand, is a positive bond.

Coming so late in Saussure’s career, subsequent to the full working out of his idea of the axis of association (now generally known as the paradigmatic axis, following Hjelmslev), it is strange to find him contemplating the possibility that a group of words have a kind of regularity that derives from their lack of regularity, their isolation, lack of attachment to a family. Negative value is perfectly Saussurean, yet the hint here that certain forms might be related associatively/paradigmatically, not by virtue of any link in sound or meaning but simply on account of being isolated within the system, is unique.

The /a/ diphthongs would be marked for rarity and isolation; and being so marked they would correlate with meanings that likewise involve marginality or abnormality. It is through the regularity of this correlation that these apparently marginal elements are incorporated into the system where everything connects. But how does this happen? Saussure’s explanation relies on another aspect of his general linguistic system, the relationship of synchrony to diachrony. He imagines

a time when there existed perhaps only four or five adjectives of ‘infirmity’ with the diphthongs ai, au, an, etc. Around this nucleus furnished by chance, ever more numerous formations will have come to fix themselves, where a certain community of ideas favoured diphthongs with a. It would thus involve a fact of lexical analogy […].

Saussure (1912: 206 [1922: 599]).

Note that he attributes the origin of this ‘nucleus’ not to sound symbolism, but to chance. Once established, however, diphthongs with /a/ were ‘favoured’ for words sharing this general idea of infirmity. The favouring would presumably take place in the competition amongst innovative forms that occurs within parole. For Saussure, the key question in language change is not ‘Why are new forms introduced?’. In parole speakers are constantly introducing new forms, of which only a tiny proportion will find the social sanction that will make them part of the langue (in a new état de langue). Rather, the key question is, ‘Why are certain forms sanctioned and not others?’ (see Saussure 1997: 47; Joseph 2012: 503). This is where the sort of analogy-driven favouring he refers to could make a difference.

The associative relations that are central to Saussure’s conception of langue make it plausible that the analogy he proposes was synchronically real for speakers — some speakers, enough for it to have left a recoverable diachronic trace, but perhaps not enough for the set of /a/ diphthongs to form a morpheme, a meaningful unit in the langue that all speakers share. Saussure had taught his students that a linguist’s work consists almost entirely of limiting what is arbitrary in language (1922 [1916]: 182). His last published paper provided an example of how to do it, in a diachronic context, by locating hidden form–meaning correlations (what Whorf would later term ‘cryptotypes’) that crystallize within parole as speakers make analogical links in their minds.

If the /a/ diphthongs had enough sound-symbolic force to leave a recoverable trace diachronically, without however forming a proper morpheme within the langue, then we are dealing with something that is both sound-symbolic and arbitrary, without there being any conflict between the two. Latin caecus had the same meaning, or signified, ‘blind’, for those speakers to whom the iconicity was ‘audible’ as for those to whom it was not. Synchronically, whether or not a speaker is aware of the correlations, the signifier signifies. That is the point of arbitrariness. It does not negate the potential force of sound symbolism in diachrony. Sound symbolism might well be part of what it is that leads a speech community to accept particular innovations rather than others. And yet, the sign still functions perfectly well as part of the language for a speaker who does not interpret it sound-symbolically. Sound-meaning iconicity does not impact upon the fundamental arbitrariness of the linguistic sign.


[1] My translations; for originals, follow hyperlinks.


Gmür, Remo. 1990. ‘Saussures “Mémoire”-Prinzipien in seinen späteren indogermanistischen Arbeiten’, in Présence de Saussure: Actes du Colloque internationale de Genève (21-23 mars 1988), edited by René Amacker & Rudolf Engler, 39-51. Genève: Droz.

Joseph, John E. 2012. Saussure. Oxford: Oxford University Press.

Magnus, Margaret. 2013. ‘A history of sound symbolism’, in The Oxford Handbook of the History of Linguistics, edited by Keith Allan, 191-208. Oxford: Oxford University Press.

Saussure, Ferdinand de. 1877. ‘Sur une classe de verbes latins en –eo’. Mémoires de la Société de Linguistique de Paris 3.279-293. (Reprinted in Saussure 1922: 353-369.)

———-. 1879. Mémoire sur le système primitif des voyelles dans les langues indo-européennes. Leipzig: printed by B. G. Teubner. (Reprinted in Saussure 1922: 1-268.)

———-. 1912. ‘Adjectifs indo-européens du type caecus “aveugle”‘, in Festschrift Vilhelm Thomsen zur Vollendung des siebzigsten Lebensjahres am 25. Januar 1912, dargebracht von Freunden und Schülern, 202-206. Leipzig: Otto Harrassowitz, 1912). (Reprinted in Saussure 1922: 595-599.)

———-. 1916. Cours de linguistique générale. Edited by Charles Bally & Albert Sechehaye, assisted by Albert Riedlinger. Lausanne & Paris: Payot. (2nd edition 1922; further editions essentially unchanged). English version, Course in general linguistics, by Wade Baskin, New York: New York: Philosophical Library, 1959 (reprinted with new introduction by Perry Meisel & Haun Saussy, New York: Columbia University Press, 2011).

———-. 1922. Recueil des publications scientifiques de Ferdinand de Saussure. [Edited by Charles Bally & Léopold Gautier.] Genève: Sonor, Lausanne: Payot, Heidelberg: C. Winter.

———-. 1995. Phonétique: Il manoscritto di Harvard Houghton Library bMS Fr 266 (8). Ed. by Maria Pia Marchese. Padova: Unipress.

———-. 1997. Deuxième cours de linguistique générale (1908–1909), d’après les cahiers d’Albert Riedlinger et Charles Patois / Saussure’s Second Course on General Linguistics (1908–1909), from the notebooks of Albert Riedlinger and Charles Patois, edited by Eisuke Komatsu, translated by George Wolf. Oxford & New York: Pergamon.

How to cite this post:

Joseph, John E. 2014. Saussure’s sound symbolism. History and Philosophy of the Language Sciences.

Tagged with: , , ,
Posted in 19th century, 20th century, Europe, History, Linguistics, Structuralism
6 comments on “Saussure’s sound symbolism
  1. Amol Padwad says:

    As you rightly point out the ‘despair’ arises out of wrongly positing sound-symbolism against arbitrariness. And your last few lines summarise a simple answer to this riddle – sound-symbolism is itself a manifestation of arbitrariness!

    • John Joseph says:

      It’s nice to be credited with finding a simple answer to a riddle; but I’m not sure what you mean by ‘manifestation’.

  2. jeanmfort says:

    Hello John,
    Thank you very much for this very interesting paper. In his 1912 paper, does Saussure positively say that some diphthongs are inherently sound-symbolic, or does he say that by the force of analogy with a semantically circumscribed group of words these diphthongs become symbolic of a definite meaning? If the latter, I think that we are not dealing with sound-symbolism but with what Bréal would call, I suppose, “fausse perception”.
    Bloomfield (in his “Introduction”, 1914: 79-80), who rejects sound-symbolism, similarly says that the feeling attached to such words as “flare” or “flash” is probably due to “nothing other than the existence of parallel words expressing the same shade of meaning with the same sounds”, and that “sound-symbolism” is in fact due to the “associative habits” of a specific community.
    JM Fortis

    • John Joseph says:

      Jean-Michel, how I wish that we had a clear definition of ‘sound symbolism’ that we all agreed on. From your brief message I can’t quite work out what it means for you, but I can tell that it’s different (narrower, I think) than my understanding of it. I gather that you’re drawing a line between the ‘associative habits of a specific community’ and the language of that community, but it’s not clear to me on what basis that line would be drawn; or on what basis one would distinguish a ‘false’ from a true perception in a case such as this; or what the origin of a ‘feeling’ has to do with the psychological reality of that feeling. Certainly Saussure could never have claimed that any diphthong or other sound is ‘inherently’ sound-symbolic, but if that leaves ‘fausse perception’ as the only alternative, there is a real gap between our understandings of these terms.

    • John Joseph says:

      Jean-Michel, I’ve thought further about your comment and think that perhaps I now have some better insight into the different uses of the terms that I mentioned initially.

      You ask: “does Saussure positively say that some diphthongs are inherently sound-symbolic?”

      Saussure hypothesises a link between the sound shape of diphthongs consisting of /a/ followed by a sonant, and the restriction of such diphthongs to a small set of Latin adjectives that all refer to some deformation or infirmity. This would be symbolic in the sense that the articulatory deviation from the ‘straight’ core vowel /a/ to the sonant is a kind of deformation. So, yes, these diphthongs would be inherently sound-symbolic in the sense that the semantic link inheres in their sound-shape in Latin. No claim is made concerning other languages.

      If though by ‘inherently’ you mean universally, such that a single word in any language that has the diphthong but not the meaning would falsify the hypothesised sound-symbolism, then I’d say that you’re recapitulating the reductio that Socrates puts to Cratylus. By that criterion, no sound- symbolism could possibly be valid. That is clearly not Saussure’s intention in the 1912 article, where, in true scientific (as opposed to philosophical) spirit, he starts with a surprising observation and invites us to consider whether it might represent something more than coincidence.

      You go on to ask: “or does he say that by the force of analogy with a semantically circumscribed group of words these diphthongs become symbolic of a definite meaning?”

      Your first question seemed to be synchronically oriented, but it isn’t clear to me whether this second one is synchronic or diachronic. Saussure speculates about how the sound-symbolism of these diphthongs might have arisen diachronically. Like all linguistic innovations, it would have begun in parole, which is to say with individual speakers and their utterances and interpretations. It happened, fortuitously, that some adjectives linked semantically by the idea of deformity or infirmity contained diphthongs with /a/, and some speakers interpreted this as not fortuitous, but symbolic. This then led to more words with this semantic feature taking on the diphthong, reinforcing perception of the symbolic link.

      That link never became part of the langue, which would have made it a fully-fledged element of Latin grammar. Yet perception of it was sufficiently widespread for it to have shaped the lexicon where this semantically circumscribed group of words was concerned. It is this perception that Saussure believed he had recovered. Synchronically, these diphthongs became symbolic of a definite meaning for some speakers — which I would not characterise as a ‘fausse perception’, nor is there any indication that Saussure regarded it as such. If there is any false perception going on, it is that a feature of a language which is psychologically real for many of its speakers is erroneous if those speakers happen not to include the ones who write the grammar books.

      Finally, the set of Latin words which Saussure identifies is not comparable to the set of English words starting with fl-, because English has hundreds of words beginning with fl-, a few small groups of which might be interpreted as sharing a semantic link (alongside flash and flare and flicker, there is flow and flood, for instance, as well as flutter and flabby and flee and flea and on and on). Saussure claims to be treating the full set of Latin adjectives with /a/ diphthongs.

      You frame the question as a dichotomy: either a sound is inherently symbolic, or else speakers have a false perception based on partial analogy. I would say that the case examined in Saussure (1912) is neither the one nor the other.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: