However, it is Ferdinand de Saussure (1857–1913) who is considered to be the founder of modern 'scientific' linguistics. Central to what de Saussure says about words are two related statements: firstly, he says that "the sign is arbitrary". This means that he considers the words that we use to indicate things and concepts could be any words - they are essentially just a consensus agreed upon by the speakers of a language and have no discernible pattern or relationship to the thing. Thus, the sounds themselves have no linguistic meaning. Secondly, he says that, because words are arbitrary, they have meaning only in relation to other words. A dog is a dog because it is not a cat or a mouse or a horse, etc. These ideas have permeated the study of words since the 19th century.
However, Saussure himself is said to have collected examples where sounds and referents were related. Ancient traditions link sounds and meaning, and some modern linguistic research does also.
Margaret Magnus is the author of a comprehensive book designed to explain phonosemantics to the lay reader - Gods of the Word. This work describes three types of sound symbol using a model first proposed by Wilhelm von Humboldt (see below):
This is the least significant type of symbolism. It is simply imitative of sounds, or suggests something that makes a sound. Some examples are: crash, bang, whoosh.
Words that share a sound sometimes have something in common. If we take for example all of the words that have no prefix or suffix and group them according to meaning, some of them will fall into a number of categories. So we find that there is a group of words beginning with /b/ are about barriers, bulges, bursting, and some other group about being banged, beaten, battered, bruised, blistered and bashed. This proportion is according to Magnus above the average for other letters.
Another hypothesis states that if a word begins with a particular phoneme, then there is likely to be a number of other words starting with that phoneme that refer to the same thing. An example given by Magnus is: if the basic word for 'house' in a given language starts with a /h/, then by clustering, disproportionately many words containing /h/ can be expected to concern housing: hut, home, hovel...
Clustering is language dependent, although closely related languages will have similar clustering relationships.
Some languages possess a category of words midway between onomatopoeia and usual words. Whereas onomatopoeia refers to the use of words to imitate actual sounds, there are languages (for example Japanese) known for having a special class of words that "imitate" soundless states or events, called phenomimes (when they describe external phenomena) and psychomimes (when they describe psychological states). On a scale that orders all words according to the correlation between their meaning and their sound, with the sound-imitating words like meow and whack at one end, and with the conventional words like water and blue at the other end, the phenomimes and the psychomimes would be somewhere in the middle (see Japanese sound symbolism). In the case of Japanese, for example, such words are learned in early childhood and are considerably more effective than usual words in conveying feelings and states of mind, or in describing states, motions, and transformations in the outside world. They are not found, however, only in children's vocabulary, but widely used in daily conversation among adults and even in more formal writing.
In the sentence 星がきらきら光っている hoshi ga kirakira hikatteiru, meaning "The stars are shining sparklingly", kirakira is a good example of a phenomime, which conveys the flickering starlight into a sequence of sounds that can be traced back to the original optical phenomenon.
The sentence 電車が込んでいていらいらしていた densha ga konde ite iraira shite ita ("The train was so crowded it was getting on my nerves.") gives an example of a psychomime: the word iraira describes the irritated state of mind due to the train's being crowded.
Several ancient traditions exist which talk about an archetypal relationship between sounds and ideas. Some of these are discussed below, but there are others as well. If we include a link between letters and ideas then the list includes the Viking Runes, the Hebrew Kabbalah, the Arab Abjad, etc.. References of this kind are very common in The Upanishads, The Nag Hammadi Library, the Celtic Book of Taliesin, as well as early Christian works, the Shinto Kototama, and Shingon Buddhism.
In Cratylus, Plato has Socrates commenting on the origins and correctness of various names and words. When Hermogenes asks if he can provide another hypothesis on how signs come into being (his own is simply 'convention'), Socrates initially suggests that they fit their referents in virtue of the sounds they are made of:
However, faced by an overwhelming number of counterexamples given by Hermogenes, Socrates has to admit that "my first notions of original names are truly wild and ridiculous".
The Upanishads contain a lot of material about sound symbolism, for instance:
The idea of phonosemantics was sporadically discussed during the Middle Ages and the Renaissance. In 1690, Locke wrote against the idea in an essay called "An Essay on Human Understanding". His argument was that if there were any connection between sounds and ideas, then we would all be speaking the same language, but this is an over-generalisation. Leibniz's book New Essays on Human Understanding published in 1765 contains a point by point critique of Locke's essay. Leibniz picks up on the generalization used by Locke and adopts a less rigid approach: clearly there is no perfect correspondence between words and things, but neither is the relationship completely arbitrary, although he seems vague about what that relationship might be.
Otto Jespersen suggests that: "Sound symbolism, we may say, makes some words more fit to survive." Dwight Bolinger of Harvard University was the primary proponent of phonosemantics through the late 1940s and the 1950s. In 1949, he published The Sign is Not Arbitrary. He concluded that morphemes cannot be defined as the minimal meaning-bearing units, in part because linguistic meaning is so ill-defined, and in part because there are obvious situations in which smaller units are meaning-bearing.
Ivan Fónagy (1963) correlates phonemes with metaphors. For example, nasal and velarized vowels are quite generally considered ‘dark’, front vowels as ‘fine’ and ‘high’. Unvoiced stops have been considered ‘thin’ by European linguists, whereas the fricatives were labelled ‘raw’ and ‘hairy’ by the Greeks.
Hans Marchand provided the first extensive list of English phonesthemes. He wrote, for example, that "/l/ at the end of a word symbolizes prolongation, continuation" or "nasals at the end of a word express continuous vibrating sounds."
Gérard Genette published the only full length history of phonosemantics, Mimologics (1976). In 450 pages, Genette details the evolution of the linguistic iconism among linguists and poets, in syntax, morphology and phonology.
Linguist Keith McCune demonstrated in his doctoral thesis that virtually every word in the Indonesian language has an iconic (phonosemantic) component. His two-volume doctoral thesis "The Internal Structure of Indonesian Roots" was completed at the University of Michigan in 1983 and published in Jakarta in 1985.
In the 2003 BBC Reith Lectures, Vilayanur S. Ramachandran outlined his research into the links between brain structure and function. In the fourth lecture of the series he describes the phenomena of synesthesia in which people experience, for example sounds in terms of colours, or sounds in terms of tastes. One type of synesthesia has people seeing numbers, letters of the alphabet, or even musical notes, as having a distinct colour. Based on his research Ramachandran proposes a model for how language might have evolved. The theory is interesting because it may explain how we make metaphors and how sounds can be metaphors for images – why for example sounds can be described as bright or dull. In explaining how language might have evolved from cross activation of adjacent areas in the brain, Ramachandran notes four crucial factors, not all related to language, but which combined might well have resulted in the emergence of it. Two of these processes are of particular interest here.
Synesthetic cross modal abstraction: i.e. we recognise properties that sounds and images have in common and abstract them to store them independently. The sounds and shapes of the objects have characteristics in common that can be abstracted, say a sharp, cutting quality of a word, and the shape it describes - what Ramachandran called the 'Bouba/kiki effect' based on the results of an experiment with two abstract shapes, one blobby and the other spiky, and asking people to relate the nonsense words bouba and kiki to them. The effect is real and observable, and repeatable.
Built in preexisting cross activation. Ramachandran points out that areas of the brain which appear to be involved in the mix-ups in synesthesia are adjacent to each other physically, and that cross-wiring, or cross activation, could explain synesthesia and our ability to make metaphors. He notes that the areas that control the muscles around the mouth are also adjacent to the visual centres, and suggests that certain words appear to make our mouth imitate the thing we are describing. Examples of this might be words like "teeny weeny", "diminutive" to describe small things; "large" or "enormous" to describe big things.
The sound of words is important in the field of poetry, and rhetoric more generally. Tools such as euphony, alliteration, and rhyme all depend on the speaker or writer confidently choosing the best-sounding word.
John Mitchell's book Euphonics: A Poet's Dictionary of Enchantments collects lists of words of similar meaning and similar sounds. For example, the entry for V begins:
Likewise, "gl-" words for shiny things: glisten, gleam, glint, glare, glam, glimmer, glaze, glass, glitz, gloss, glory, glow, and glitter. In German, nouns starting with "kno-" and "knö-" are mostly small and round: Knoblauch "garlic", Knöchel "ankle", Knödel "dumpling", Knolle "tuber", Knopf "button", Knorren "knot (in a tree)", Knospe "bud (of a plant)", Knoten "knot (in string or rope)".