The "arbitrariness of the sign" is a fundamental principle of modern linguistics: that is, that there's nothing intrinsic about the sound sequence [kʰæt̚] or the phoneme sequence /kæt/ that links it to a small furry animal. English-speaking humans have just decided arbitrarily that the animal should be a /kæt/, and parents teach it to their children, and thus a /kæt/ it remains.

However, this principle isn't necessarily obvious. Plato, among others, wrote long philosophical tracts on why specific sounds are linked to specific meanings. And every so often a new theory pops up (sometimes aiming to be scientific, sometimes not) proposing the same thing. ("Edenics" is the first one that comes to mind, but there are many, many others out there.)

So—what is the current evidence for the arbitrariness of the sign? In other words, if someone comes to me asking how I know that [kʰæt̚] isn't intrinsically linked to 🐈, what should I point them toward?

(The most obvious answer is "because some people call 🐈 /neko/, and others call it /paka/, and neither of those people would necessarily understand what /kæt/ means"—but it's very easy to find coincidental resemblances between languages when your data set is large enough, and these theories usually have dozens and dozens of them. I'm looking for an argument that doesn't depend on specific examples, since they can always come up with more coincidences in response.)

3 Answers 3


Before you point someone to evidence, I suggest pointing them to the problem of saying exactly what the "principle" or at least empirical claim is. It is about the relationship between meaning and pronunciation, across languages. In English, the relationship between form and meaning is not arbitrary, it is "natural" – it is in the nature of contemporary English that the word for "cat" is indeed approximately /kæt/, and it is not /dɔg/.

If there were a natural relationship between form and meaning across languages, we would expect there to be significant correlations between form and meaning, e.g. words for "dog" would tend to look like [dɔg], or perhaps [mbwa]. Hmmmmm, bad example, the latter is kind of true (hundreds of languages have [mbwa] as the word for "dog", nearly a thousand languages have a word like [masa] as the word for "eye", virtually all languages have the same word for "computer" – for known reasons). We'd expect there to be a correlation between the meaning "small" and [titi] (rats, I gotta get a better example). As you know, it is not arbitrary that the modern English word for "cat" is /kæt/. It's pronounced that way because our ancestors pronounced it that way. [mbwa] is one of those Ur-onomatopoetic words, idem [titi]. If you don't define the claim clearly enough, you can never escape the problem that there are lots of non-random sound-meaning relations.

You might profitably point to the more random nature of the relationship between meaning and form with grammatical morphemes. For example, collect all of the noun plural morphemes of languages, the nominative singular markers, the past tense markers... then see if anyone can correctly reassign function based on form (even if you tell them what the possibilities are).


Following Max Müller in his Lectures on the Science of Language, one may study the etymologies of supposed onomatopoetic words to see whether they obey the sound laws that have governed the evolution of the bulk of vocabulary, or whether words that sound like their referents are exceptions to the sound laws. Mostly, they are not exceptions, though there are a few cases in which they are.

Müller also gives cases where words heralded as due to the "onomatopoetic force" have come to this apparent status only through the action of the sound laws, which obviously in their application to most vocabulary have nothing to do with onomatopoeia. So etymology shows that the onomatopoeia was coincidental.

  • It seems like these arguments wouldn't go against a hypothesis that sounds laws affect onomatopoetic words like other words, but the use of the forms that result from sound change is affected by how they sound (for example, there might be shifts in meaning or shifts in frequency of usage after a sound change) Aug 15, 2019 at 19:04
  • @sumelic, yes, I agree with your comment.
    – Greg Lee
    Aug 15, 2019 at 21:21
  • Onomatopoea are to a degree arbitrary, because of the limitations of the vocal tract. In consequence, they don't immediately gain in arbitrariness after alterations, even if the sound change were random. But Onomatopoea are inherently not completely arbitrary, if that's why you chose them as example to begin with! Or would you say onomatopoea are arbitrary at their conception? The point at which they become lexified is probably when imitations are imperfectly imitated. yada yada yada
    – vectory
    Jan 7, 2020 at 22:20

Reading my answer to the question is best preceded by reading these texts first because I will reference to them with examples.



I am excluding onomatopoeia in this answer.

When you have read all the views on arbitrary language (the pratclif link), eventually the author writes: “Some signs are absolutely arbitrary; in others we note not its complete absence but the presence of degrees of arbitrariness: the sign may be relatively motivated, for example, 'vingt' is relatively less motivated than 'dix-neuf' (which by its form refers to other forms) ... but even in the most favourable cases motivation is never absolute ...” So if we don’t know or can not find a motivation, the word must be arbitrary? That is like saying I am blind so light can not exist. What if all words have a motivation? Sure we may have lost the motivations of most of the words in languages, but that does not mean they don’t exist. Here is where people ‘believe’ my methods are based on coincidence, nonsense, twaddle and whatever terminology can be used to keep up the misconception that words are arbitrary. If we don’t see words as combined phonetic elements with descriptive meaning we can not analyse if there is a natural origin or motivation for each sound/syllable within the word connected to meanings. I write ‘meanings’because I suspect multiple basic meanings for each sound. So stating that most words are arbitrary is at least premature. And I must confess that the evidence for my motivations is empirical. We are humans and language is a skill in which we stand out. We give meaning to signs. And if we intend meaning we intend motivation.

So, what is the challenge we have? What if there are absolute motivations for each word? What if people just lack the ability to identify them all? This inability to identify them puts the mass of valuable data and possible outcomes in the bin, because it is not understood as the seminal thought author points out. However his example ‘vingt’ (20) can similarly be motivated as dix-neuf (19 literally ten-nine) referring to two other forms. Vingt is a corrupted compound of Latin ‘quinque quattuor’ which forms show the initial speaker’s intention of conveying a multiplication: 5 x 4 to be 20. This is common in French counting because 80 is expressed as quatrevingt which literally says fourtwenty (4x20=80 so not a sum but another motivated multiplication). An English speaker not knowing this difference in counting can easily claim ‘vingt’ to be a more or less arbitrary word.

If we leave out the culture in the analyses, we are again blind for what use the word has in a language. If we include culture we can motivate why ‘dog’ is based on observing the behaviour of the animal ‘digging’ holes, while ‘hound’ can be motivated by the animal’s use in ‘hunting’ and ‘perro’ is the guardian of the ‘house’ just like ‘Canis’ can be motivated by the animal’s use in protection by its ‘teeth’, while ‘kalba’ is motivated by observing the animal produce ‘saliva’: drewl. The treatment of dogs within the cultures using these languages coincides with the level of usefulness to humans. In northern Europe the dog was used in hunting and thus providing important food especially in the winter. The dog culturally became a friend of humans. In Arabic countries living a dog’s life is not easy and this can be understood by its name; his drewling not being a useful function for humans. It was not used as extensively in hunting as in the north. And in Southern Europe the animal was used less in hunting but more in protection of the house and land. Hence a motivated connection with teeth and house.

A noun is a descriptive name often based on a quality of the context it signifies. And from that description it is possible to analyse older meanings for the single phonemes by comparing their use across languages.

I found that in husbandry (motivated by compounding the descriptive location of the animals: house & boundary) naming of animals follows certain specific rules. The word for a 🐈 being ‘cat’ in English is not arbitrary, neither natural. It is cultural. There is no direct link between the form and the meaning of cat. There is a direct link though between the name of the animal and its use for humans. A cat/Katze catches mice. The motivation of the word ‘cat’ is found in the cultural context of the usefulness of the cat catching mice. Cat is than cognate to ‘chase’, French ‘chasse’ (hunt) and Spanish ‘caza’ (hunting/chase) and even ‘casa’ (house), which in extension encloses the sleeping place (instead of rodents).

You asked about one example where a sound carries a specific meaning. This essentially is the biggest question of all. How did the first sounds link to the first meanings? If we consider the development of language as a string of DNA, we have to analyse each word to its origin until there are only single sounds to designate meaning to. That is painstaking work. Something that should be done by computers but as it is a hobby of mine I have done a bit myself. I am not ready but can claim a handful. There are 2 that I suspect are the sounds copied from our environment and they fortunately share a context which makes the case a bit stronger. This does not mean that it is the only natural meaning to the sound, because there are many linguistic developments that give more possibilities. These sounds are /v/ (sometimes /f/) and the /n/.

The /v/ dominate words that have to do with flying in Germanic languages. In Semitic it is found in the context of heaven in Hawwah (Biblical Eve) which means ‘snake’ and ‘sky’. Two meanings related by the sound of air if directed by blowing through the mouth. French ‘vent’ and English ‘wind’ (sound change). So whistling and blowing air most probably became the natural origin for the sound /v/ in parole and language. And if used in descriptions that compounded to names and verbs, this would explain the relation between a phoneme in a word and similar meanings within different contexts.

/n/ is the nasal sound. Based on the evidence above one would expect the /n/ to dominate the nouns meaning ‘nose’ in different languages which it still does. It is found in the Arabic word for breath (a function of the nose) and English ‘sneeze’. The verbs ‘to smell’ and ‘to smoke’ use different phonemes that can be found in the Hebrew word for sky (Shamash). The letter Samek is said to derive from a support, thus lifting something. The skyward context can be noted again. It is very hard to motivate all words with an /n/, but the words ‘no’ and ‘not’ (nicht in German) are based on ‘night’. Yes is harder to motivate because the connection with ‘day’ is harder to see at first sight. But that is the context where we have to look. The ‘eyes’ and ‘yes’ are cognate. Compare: French: oui and ouil Greek: μάτι versus French matin Greek: όχι versus oog (eye in Dutch) But in Greek the sounds for these meanings seem switched. Greek όχι (no) seems cognate with ‘oké’ which is in itself strange because they are antonyms and we could associate Yes with the presence of light and no with the absence of light (night). Greek Yes is Ναί. A choice that amazes me. But it shows how cultures invent or invert or substitute meanings, contexts, forms and everything out of unknown reasons (word play, political, religious or other). It is just one of many underlying ways words and sounds are related to meaning. So like I said painstaking.

What is the evidence for the arbitrairiness of a sign?

The answer must be, the lack of or inability to find motivations for words through qualities of the concept.

Add to that unwillingness (people calling my research outcomes coincidence, nonsense, twaddle and whatever terminology that does not support rational, logical and argumented reasoning and evades dialogue.

On a personal note: Draconis, I Thank you for your attitude which has always been informative, critical and investigative. You have been an island of tolerance here. And your question motivated me to determine where I stand in the realm of linguistic research by comparing my ideas to others. Thanks also to anyone editing this as I am not a native speaker of English.

  • 3
    Tolerance or no tolerance, as in your other questions and answers, your insistence on using demonstrably incorrect etymologies mars whatever point you intend to make... Aug 17, 2019 at 15:02
  • Thanks Luke but you are not contributing to the question. Mention which etymologies and why. I don’t base my research on written evidence because its flaws are not recognized by the linguists who invented the rules. Talking about arbitrary... If you have a better answer, please contribute. Else it is just your opinion.
    – Ajagar
    Aug 17, 2019 at 16:51
  • Comments are a perfectly good place to state one's opinion about the worthiness and scientific value of an answer. So is, of course, a -1.
    – LjL
    Aug 17, 2019 at 17:55
  • "for example, 'vingt' is relatively less motivated than 'dix-neuf' (which by its form refers to other forms) ... but even in the most favourable cases motivation is never absolute" I think you're misunderstanding what this quote is saying. French vingt isn't entirely arbitrary: it comes from Latin vigintī, which comes from a reconstructed PIE form meaning "two tens". What they're saying is that it might as well be arbitrary to a French-speaker, because it has no connection to "two tens" within the modern French language.
    – Draconis
    Aug 17, 2019 at 19:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.