Arbitrariness of the Linguistic Sign

The principle and where it came from

Ferdinand de Saussure (1857–1913) was a Swiss linguist who, after early work in Indo-European reconstruction, taught a series of three courses on general linguistics at the University of Geneva between 1907 and 1911. He died in 1913 having published almost nothing on the subject. Two of his students, Charles Bally and Albert Sechehaye, edited their own and other students' lecture notes into the Cours de linguistique générale, published in 1916. The book founded structural linguistics; the "arbitrariness of the linguistic sign" is its first analytic claim.

Saussure's argument runs like this. A linguistic sign has two faces: a signifiant (signifier — the sound shape, or strictly its psychological imprint) and a signifié (signified — the concept). Neither face has any necessary relation to the other. The proof is comparative: every language attaches a different sound shape to roughly the same concepts, and no language is more correct than another. The bond between "tree" and the concept of a perennial woody plant is fixed by community convention, not by any tree-like property of the sounds /tri:/.

Saussure was careful about what arbitrariness does not mean. It does not mean a speaker can choose any sound for any meaning — once a convention is established, the speaker is bound by it. It does not mean signs are random or unsystematic — within a language, signs form a tightly structured system whose values depend on their oppositions to other signs. Arbitrariness is a property of the form-meaning bond, not of the use of language.

Absolute versus relative arbitrariness

Saussure himself qualified the principle. He distinguished:

• Absolute arbitrariness. A simple, monomorphemic sign like "tree" or "dog" — no internal structure relates the sound to the meaning.
• Relative arbitrariness. A complex sign whose meaning is partly motivated by the language's own system. "Nineteen" is built from "nine" plus "-teen," making it relatively motivated by speakers who already know its parts. The roots remain absolutely arbitrary, but the construction is transparent.

Languages vary in how much relative motivation they exhibit. Chinese 火车 (huǒchē, "fire-vehicle" = train) is relatively motivated by its components; English "train" is absolutely arbitrary. Sanskrit verbal morphology is highly motivated within the system. The grade of relative arbitrariness differs cross-linguistically and even across vocabulary domains within a single language.

Arbitrariness compared to motivation

Sign type	Form–meaning relation	Example	Saussurean status
Symbol (linguistic sign)	Arbitrary, conventional	English "tree"	The default; first principle of language
Icon	Resemblance	A portrait, a road sign for a curve	Non-linguistic; Peirce's category
Index	Causal or contiguous	Smoke means fire; deictic "here"	Largely non-linguistic; deixis is a partial exception
Onomatopoeia	Partial sound resemblance	"buzz," "crash," "meow"	Marginal in the lexicon, still language-specific
Sound symbolism	Cross-linguistic statistical bias	Bouba (rounded) vs. kiki (sharp); /i/ for small things	Modest but reproducible — softens strong arbitrariness
Ideophone	Depictive iconic word class	Japanese pikapika "twinkling"; Siwu kpɔlɔŋkpɔlɔŋ "rolling"	Productive iconicity in many languages — major counterevidence
Compound (relative arbitrariness)	Motivated by parts	"blackboard," "nineteen," 火车	Saussure's second-grade arbitrariness

The columns show that arbitrariness is best understood as a gradient, not a binary. Saussure's strongest claim — that all linguistic signs are arbitrary — is too strong; his more measured claim — that arbitrariness is the dominant principle, with iconic exceptions — survives modern scrutiny.

Worked example: how diversity proves the principle

Take the same domestic pet — a dog — across languages:

• English: dog (/dɔɡ/)
• French: chien (/ʃjɛ̃/)
• Spanish: perro (/ˈpero/)
• German: Hund (/hʊnt/)
• Japanese: 犬 inu (/inu/)
• Mandarin: 狗 gǒu (/kou/)
• Arabic: كلب kalb (/kalb/)
• Swahili: mbwa (/mbwa/)

None of these sound shapes is more dog-like than another. None contains an inherent connection to the four-legged barking creature. The fact that they all label the same concept while sounding nothing alike is the empirical proof of arbitrariness. If signs were motivated by their referents, we would expect convergent forms; what we observe is divergent forms held together only by community convention.

Even within one language, etymologies show how arbitrary the link is. English "dog" replaced Old English "hund" (cognate to German Hund) starting around the 14th century — the etymology of "dog" itself is uncertain. Replacement is possible because the sign was never tied to its meaning by anything stronger than convention.

Counterarguments and modern revisions

Three lines of evidence have softened the strict Saussurean position:

• Sound symbolism (bouba-kiki). Wolfgang Köhler (1929) found that subjects systematically associate the nonsense word "baluba" with rounded shapes and "takete" with jagged shapes. Ramachandran and Hubbard (2001) replicated with bouba/kiki across English, Tamil, and Himba. The effect is small but real — about 95% agreement in some studies — and demonstrates non-arbitrary form-meaning mappings present even in nonsense words.
• Statistical iconicity. Blasi, Wichmann, Hammarström, Stadler, and Christiansen (PNAS, 2016) analyzed 100 basic concepts across 4,298 languages and found systematic biases — "nose" tends to contain /n/, "tongue" tends to contain /l/, "small" tends to contain /i/. The biases are subtle but pervasive, suggesting that language is more iconic than Saussure allowed.
• Ideophones. Mark Dingemanse and colleagues have documented iconic word classes in Japanese, Korean, Yoruba, Siwu, Ewe, and many other languages. These are not marginal — Japanese has thousands of ideophones in productive use. Their existence shows that iconicity can be a structural part of a language, not just a marginal exception.

Modern semioticians (Hilpert, Dingemanse, Perniss, Vigliocco) propose a continuum: most signs are arbitrary, but iconicity is widespread and useful, especially in early acquisition and in expressive registers. The Saussurean first principle is preserved but qualified — arbitrariness is the default, not the exception, but it is far from absolute.

Variants and related principles

• Conventionality. Closely related to arbitrariness — the sign means what it does because the speech community has agreed it does. Conventionality persists even when iconicity is high (an ideophone is still community-fixed in form).
• Discreteness. Hockett's feature that sound categories are categorical, not continuous; complements arbitrariness by allowing a finite set of signs to be reused without confusion.
• Duality of patterning. A small set of meaningless sounds combine into a large set of meaningful units. This combinatorial layer is what allows arbitrary form to map to a vast number of meanings.
• Iconicity in sign languages. Sign languages are visually rich and contain more iconicity than spoken languages; many signs depict their referents (a hand for "tree," a face for "cry"). Yet Klima and Bellugi (1979) showed that historical change in sign languages drifts toward arbitrariness — iconic origins are eroded by phonological evolution.
• Saussure's second principle. The linearity of the signifier — that speech sounds unfold in time, in a single dimension. Arbitrariness and linearity together define the structural sign.

Common pitfalls

• Equating arbitrariness with randomness. Arbitrary signs still form a system whose values are fixed by their oppositions. Convention is not chaos.
• Treating onomatopoeia as a counterexample to all arbitrariness. A few iconic words do not refute the principle for the rest of the lexicon.
• Assuming arbitrariness is total. Modern work on sound symbolism shows the principle is statistical, not absolute.
• Confusing arbitrariness with the speaker's freedom. Saussure was emphatic: speakers cannot change signs at will. Arbitrariness is a structural property, not a license.
• Ignoring relative arbitrariness. Compounds and derivations are partially motivated; missing this layer flattens the typology of signs.