In Plato's Cratylus, Socrates rejects the notion that the names of things have a divine origin. Language, as sign, is inferior to the thing signified. This imperfection could be used to account for the variety of languages, the multiplicity of systems of signification used to talk about, assumedly, one reality. Yet such a position suggests that our names are derivative of things - or, to put it another way, that the singular reality which we experience, filled with a variety of phenomena which we ostensibly share with others, provides a kind of prototype from which language is drawn. There is, however, another theory about language and its connection with the things around us: that the language we use actually shapes our phenomenal world, and that the words we use have the power to alter our perception of reality. To some degree it may even be arguable that the use of a particular language - or even the use of a particular vocabulary or dialect within a language - can actually define the phenomenal contents of our world. I believe that the way in which language shapes, directs, and forms our thoughts and our perception of the world suggests that realism - which would attempt to tie our words and associated conceptions to universal, unchanging forms or essences - overestimates the universality of human concepts and underestimates the power of names to shape our reality.
Realism's variety and tradition is far too diverse and expansive to receive a full treatment. Indeed, there are many subtle varieties of realism that may avoid the criticism which I will present. For my purposes, however, I will define realism as consisting primarily in the belief that things are what they are because of their ontological relationship to an eternal, unchanging reality called an essence. This ontological relationship is described in many ways, but the main vocabulary I will use is participation. An entity - physical or non-physical - participates in a form, which is eternal and unchanging. When we perceive such an entity, our intellects are able to abstract this form from the particulars of the phenomena around us: we are able to see past certain accidental or non-essential qualities, such as the fact that the thing is sitting in water, is brown, is chewing on a fish, and so forth, and are able to abstract that the entity is, in essence, a duck.
Words, then, are conceived of as socially agreed upon utterances signifying these essential concepts. Whatever variety there is in language does not change the fundamental content of these universal concepts: a man speaking Mandarin Chinese would possess the knowledge of the same form, "duckness," as an English speaker, even if the utterances used to signify that concept differ.
Language, on the realist's account, is always primarily derivative, in the sense that the content of our mental concepts shapes our language rather than vice versa. This theory nicely fits with any philosophy that requires the universality of certain sorts of concept: for example, natural law, among others.
However, if language is not primarily derivative, and if instead the utterances we use actually alter the content of our mental concepts - that is, if language changes the way in which we perceive the world - then realism's claims that humans have access to universally uniform mental concepts in the form of essences is doubtful.
Indeed, research on this subject is emerging which reveals that language shapes the content of our mental concepts as well as our phenomenal world. Take, for example, the experiments conducted by Lera Boroditsky of Stanford University. Comparing the speakers of a language like Kuuk Thayorre, a language that insists so much on absolute frames of reference that it lacks concepts such as "left" and "right" and instead requires all statements to be made in terms of the cardinal directions, to speakers of a language like English, which tends to use more relative spatial terms, Boroditsky found that the language spoken by the individual had profound effects on his spatial and temporal perceptions. Kuuk Thayorre possess a nearly perfect sense of direction in any environment, even indoors, and tend to see temporality in terms of cardinal direction, especially east to west. When asked to arrange photographs in temporal order, Kuuk Thayorre speakers place the cards in line from east to west; English speakers, not surprisingly, will place the cards in order from left to right, while possessing none of the unique spacial perceptive abilities of the Kuuk Thayorre speakers.
Furthermore, her research revealed that the choice of conceptual metaphors and idioms - especially temporal idioms - alters a speaker's perception of the passing of time. English speakers, who tend to use idioms of length to speak about the passage of time, actually perceive the passage of time differently when thinking about distances than when not thinking about distances, while Spanish speakers, who tend to think of time in terms of quantity or size, perceive the passage of time as longer when thinking about quantity and size.
These differences in language - and the resulting differences in how we actually perceive the world - go on and on. Speakers of languages with a greater number and variety of words for color are actually better able to distinguish different shades of colors; the different genders of words in different languages results in differences in speakers' perceptions of those objects; the same object will be described by vastly different terms in different languages; and so forth. What is, in a sense, another important distinction is that these differences in perception will carry over even if a speaker is speaking in a different language: that is, a native Spanish speaker speaking English and a native German speaker speaking English will use an entirely different English vocabulary - and, sometimes, even contradictory vocabulary - to describe the same object based upon the gender biases of their native language.
These findings all serve to support the conclusion that language is not merely derivative in nature: that, rather than our utterances being simply socially determined representatives of otherwise uniform mental concepts, the particular characteristics of our language serves to inform and determine the content of our mental concepts. Of course, this does not mean that our concepts are wholly products of language - we know that animals without complex languages, such as monkeys, are able to perform tasks that demonstrate the use of mental concepts, including number. However, in studies of various people groups in South America who lack numeral language, researchers have discovered that the lack of numbers in their spoken language correlates to numerical imprecision: that is, when asked to do rudimentary numeric exercises such as identifying how many of something there were or in the performance of basic arithmetic, these people groups were consistently unable to provide exact numbers, even if their estimates demonstrated an ability to handle numerical concepts in a rough or ready way. This latter experiment suggests that even if language does not play the sole role in shaping our mental concepts, it does play a role in refining those concepts; when combined with the earlier experiments, I believe it is possible to see that the content and precision of our mental concepts is not indifferent to the utterances used to describe them.
If, then, the content and precision of mental concepts is partially derived from language, we have reason to at least refine the realist notion that utterances are merely social constructs depicting otherwise uniform, universal, or eternal concepts. It is noteworthy that nominalism would not suffer the same weakness from this finding: indeed, the nominalist account of utterances (or names) as sets of similar phenomena described under common words can very readily accept that the act of naming is itself a variable determining the content and precision of the mental concepts which we name. On a broader scope, this also suggests that attempts at formulating universal definitions or universal systems - that is, any attempt to describe a set of universal human concepts - must account for the linguistically (and, we might add, culturally, socially, and historically) conditioned nature of these concepts.