A woman operates an antique phonograph next to a man wearing a traditional Native American headdress in a sepia-toned photograph.

The ethnologist Frances Densmore, working for the Bureau of American Ethnology, and Mountain Chief, a Blackfoot leader. Photo from 1916. Courtesy the Library of Congress


Our language, our world

Linguistic relativity holds that your worldview is structured by the language you speak. Is it true? History shines a light

by James McElvenny + BIO

The ethnologist Frances Densmore, working for the Bureau of American Ethnology, and Mountain Chief, a Blackfoot leader. Photo from 1916. Courtesy the Library of Congress

Anyone who has learned a second language will have made an exhilarating (and yet somehow unsettling) discovery: there is never a one-to-one correspondence in meaning between the words and phrases of one language and another. Even the most banal expressions have a slightly different sense, issuing from a network of attitudes and ideas unique to each language. Switching between languages, we may feel as if we are stepping from one world into another. Each language seemingly compels us to talk in a certain way and to see things from a particular perspective. But is this just an illusion? Does each language really embody a different worldview, or even dictate specific patterns of thought to its speakers?

In the modern academic context, such questions are usually treated under the rubrics of ‘linguistic relativity’ or the ‘Sapir-Whorf hypothesis’. Contemporary research is focused on pinning down these questions, on trying to formulate them in rigorous terms that can be tested empirically. But current notions concerning connections between language, mind and worldview have a long history, spanning several intellectual epochs, each with their own preoccupations. Running through this history is a recurring scepticism surrounding linguistic relativity, engendered not only by the difficulties of pinning it down, but by a deep-seated ambivalence about the assumptions and implications of relativistic doctrines.

There is quite a bit at stake in entertaining the possibility of linguistic relativity – it impinges directly on our understanding of the nature of human language. A long-held assumption in Western philosophy, classically formulated in the work of Aristotle, maintains that words are mere labels we apply to existing ideas in order to share those ideas with others. But linguistic relativity makes language an active force in shaping our thoughts. Furthermore, if we permit fundamental variation between languages and their presumably entangled worldviews, we are confronted with difficult questions about the constitution of our common humanity. Could it be that there are unbridgeable gulfs in thinking and perception between groups of people speaking different languages?

The roots of our present ideas about linguistic relativity extend at least as far back as the Enlightenment of the late 17th to the 18th century. Enlightenment discussions were often couched in terms of the ‘genius’ of a language, an expression first coined in French as le génie de la langue. The term was used in a wide variety of senses, to the point where it was often not clear what precisely was meant. One contemporary commentator remarked: ‘[W]e often ask what is the genius of a language, and it is difficult to say.’ What we can say is that the genius of a language was understood as representing its distinct character, the je ne sais quoi that constitutes the idiomatic in each idiom. This unique character was frequently taken to embody something of the national mentality of the speakers of a language.

A classic – and highly influential – formulation came in 1772 with the Treatise on the Origin of Language by the German philosopher and poet Johann Gottfried von Herder (1744-1803). In opposition to contemporaries who saw the ultimate origins of human language in animal cries, Herder insisted that there is a difference in kind between human and animal communication. Human language, so Herder argued, rests on the irreducible human capacity for ‘reflection’ (Besonnenheit), our ability to recognise and think about our own thoughts. In coining our words, we reflect on the properties of the things they name, and choose the most salient of these. Different peoples will have focused on different properties, with the result that each language with its characteristic forms will encapsulate a slightly different perspective on the world. As languages are passed on from generation to generation, the differences between them accumulate, making the languages and the worldviews they contain more and more distinct. In order to understand the unique perspective of each language, we must trace the forms of words back to their etymological origins.

The Herderian thread was picked up in the early 19th century and woven most expertly into a broader account of language and literature by Wilhelm von Humboldt (1767-1835). Humboldt endorsed an element of linguistic determinism – that is, that language not only reflects a particular worldview but is actively involved in shaping it: ‘Language,’ he wrote , ‘is the forming organ of thought.’ The relationship he envisaged, however, was not one-way but dialectic. Between language and thought there inheres an endless feedback loop: our thoughts shape our words, and our words shape our thoughts. His account was not restricted to individual words – more important were the grammatical structures exhibited in the languages of the world. But even the study of grammar was only a preliminary to the real task, according to Humboldt. Grammar and vocabulary merely represent the ‘dead skeleton’ of a language. To capture its character, to see its ‘living structure’, we must appreciate its literature, the use made of the language by its most eloquent speakers and writers.

The inner form of a language, Steinthal believed, was the perfect window to the national mind

Despite Humboldt’s exhortations to seek the life of language in literature, his successors in the 19th century concentrated on devising classifications of languages revolving around their grammatical features. The goal was often described as identifying the ‘inner form’ of each language. ‘Inner form’ was a term used by Humboldt (although only fleetingly) to refer to the underlying structure and organisation of a language, as opposed to its ‘outer form’, the externally perceptible features of its words, grammar and sound system. Humboldt’s inner form carries forward the concerns of the Enlightenment’s genius of a language, while the outer form consists of the pedantic details of noun declensions, verb conjugations, regular sound substitutions and so on.

Many scholars working in Humboldt’s wake adopted his ‘inner form’ and developed it in different directions, although the most prominent version of this notion was that elaborated by Heymann Steinthal (1823-99). Inner form served as the centrepiece of Steinthal’s classification of languages, which in turn lay at the heart of his Völkerpsychologie, the ‘psychology of peoples’ or ‘ethnopsychology’. The overarching aim of Völkerpsychologie was to describe the supposed shared mentality of each nation. The inner form of a language, so believed Steinthal, was the perfect window to this national mind.

But during the course of the 19th century, talk of national minds and the character of languages fell out of fashion in the academic study of language. In this period, comparative-historical grammar became established as the premier field of linguistics. This is the approach that carefully compares words and grammatical forms across languages in order to chart their historical changes and identify their putative genealogical relations. Comparative-historical linguistics tells us, for example, that French, Italian and Spanish are all descended from Latin; that Hindi-Urdu, Bengali and Punjabi can trace their ancestry to Sanskrit; and that all these languages, along with many others traditionally spoken from western Europe to northern India, are part of the extended Indo-European family.

The hypothetical progenitor of this great family, Proto-Indo-European, has been lost to time, but elements of its vocabulary, grammar and sound system can be reconstructed from the traits of its descendants. Crucially, these are all aspects of the ‘outer form’ of languages – and the linguists who investigated these outer forms preferred to describe the historical transmutations they studied in terms of ‘sound laws’. Sound laws are mere statements of fact, that a sound attested in a certain phonetic environment in a parent language changes into other sounds in its descendants. Such accounts avoid invoking any hidden, underlying explanatory principles. Most comparative-historical grammarians believed that, for linguistics to be considered a serious science, it must limit itself to solid, objectively observable data. Uncovering the inner life of languages, capturing their characters and connections to thought and culture, were at best seen as future tasks for a thoroughly grounded science of language. At worst, they were taken to be nothing more than idle metaphysical speculation.

In what was the last gasp of the Humboldtian tradition in academic linguistics of the 19th century, the sinologist and general linguist Georg von der Gabelentz (1840-93) proposed a new subfield of ‘typology’, which would exhaustively survey the grammatical features of the world’s languages in order to discover the ‘typical traits, the ruling tendencies’ that determine linguistic structure. This undertaking would provide an empirical foundation for the ‘highest task’ of linguistics, explaining such structural tendencies as manifestations of the national mind. Gabelentz’s call for the new field fell on deaf ears in this age dominated by historical-comparative grammar. Typology would re-emerge as a mainstream pursuit in linguistics only in the early 20th century.

In this same period, on the other side of the Atlantic, questions of mind and language did enjoy currency in a Humboldtian-inflected anthropology. Franz Boas (1858-1942), the ‘father’ of American anthropology, set out to compile the definitive compendium of the Indigenous languages of North America in his multi-volume Handbook of American Indian Languages, the first volume of which appeared in 1911. The grammatical descriptions contained in Boas’s handbook were to ‘depend entirely upon the inner form of each language’. ‘In other words,’ Boas elaborated, ‘the grammar has been treated as though an intelligent Indian was going to develop the forms of his own thoughts by an analysis of his own form of speech.’

But Boas was beset by an ambivalence about the implications of the mind-language nexus. Nineteenth-century discourse on the differences between nations was all too often predicated on an assumed hierarchy of humanity. There was a widespread belief that the peoples extant in the world today had reached different stages of evolution in their societies and cultures – and that this was attributable to differences in their cognitive abilities. At the top of the hierarchy was 19th-century European man, who had unfolded his mental powers in all directions, while at the bottom were the various Indigenous peoples of the world, usually believed to be stuck in an eternal childhood of humanity or to have degenerated from a previous state of ‘civilisation’.

Attitudes were not monolithic: there were many different schemes of human social, cultural and cognitive evolution in this period, admitting of many nuances. But even such figures as Humboldt, Steinthal and Gabelentz, who revelled in human diversity and praised the uniqueness of each language, were more partial to some languages than to others. American languages, argued Steinthal, actually have no inner form. The indisputably complex constructions attested in their grammars are merely mash-ups of concrete conceptual material without any underlying formal structure. Attitudes at the time among the leading anthropologists and linguists in the United States were even more extreme.

Anthropologists continued to consider possible connections between language and mind

Boas pushed back against such prejudiced schemes. He actually agreed with his opponents about the existence of some alleged deficits in Indigenous languages, but refused to see these as an index of mental development. Many American languages lack abstract terms and indefinitely large numbers, conceded Boas, but this is not because their speakers are incapable of grasping such concepts. It is simply the case that they have never had need to talk in abstract terms or count to higher numbers, and so have never had occasion to produce such forms in their languages. If this need arose, their languages would soon adapt.

Boas’s views were largely inspired by the teachings of his former mentor in Berlin, the ethnographer Adolf Bastian (1826-1905). Bastian advocated the principle of the ‘psychic unity of mankind’, the idea that all humans, no matter what their ancestry or their present cultural condition, have at base the same mental faculties and abilities. The apparently differing ‘ethnic thoughts’ of the various peoples of the world are merely different arrangements of the same ‘elementary thoughts’ common to all humanity. The human mind is essentially the same everywhere.

We therefore see in the 19th century a clear arc in the development of academic attitudes towards linguistic relativity. At the beginning of the century, linguistic relativity was a respectable position in the study of language, buoyed by the writings of such figures as Herder and Humboldt. But as the century wore on, academic linguistics became increasingly dominated by the school of comparative-historical grammarians, whose approach was highly technical and empirical in character. In this intellectual environment, linguists gradually turned away from the seemingly nebulous questions about the underlying conceptual apparatuses of languages. Anthropologists, by contrast, continued throughout the 19th century to consider possible connections between language and mind, but the hierarchical terms in which their discussions were often framed came under criticism towards the end of the century, in a movement spearheaded by Boas.

The latter-day ‘Sapir-Whorf hypothesis’ is in many respects a continuation of the 19th-century debates. Edward Sapir (1884-1939) and his student Benjamin Lee Whorf (1897-1941) were heirs to the Humboldtian tradition. Sapir was steeped in German language scholarship: his Master’s thesis was on Herder’s Treatise on the Origin of Language. He was also one of Boas’s most talented and devoted pupils, and perpetuated his teacher’s positions. ‘Language and our thought-grooves are inextricably interwoven,’ wrote Sapir in 1921, ‘are, in a sense, one and the same.’ But, like Boas, he insisted that there are no ‘significant racial differences’ in thought across the human species, and no direct connections between culture and language. It is therefore impossible to infer alleged evolutionary stages from language structure: ‘When it comes to linguistic form, Plato walks with the Macedonian swineherd, Confucius with the head-hunting savage of Assam.’

Despite the desire to extricate his research from the prejudices of past scholarship, Sapir was still invested in the project of analysing the grammatical ‘processes’ and ‘concepts’ attested in the languages of the world in order to identify the ‘type or plan or structural “genius”’ of each language. But this endeavour was tempered by a belief in the at least partial autonomy of linguistic form. On Sapir’s account, every language possesses an ‘inner phonetic system’ and ‘a definite feeling for patterning on the level of grammatical formation’, both of which ‘operate as such, regardless of the need for expressing particular concepts or of giving external shape to particular groups of concepts.’ Language, it would seem, was not quite so stuck in those thought-grooves.

Treating linguistic form as in some way autonomous was implicit in the 19th-century comparative-historical grammarians’ postulation of sound laws. In the 20th century, an explicit move was made by many linguists to hive off language structure as their private domain, an object they could investigate independently of any broader questions of cognition or the physical production and reception of speech. In these years, the Genevan linguist Ferdinand de Saussure (1857-1913) introduced a distinction between la langue (language) and la parole (speech), a distinction that has become fundamental to much subsequent linguistic scholarship. La langue is the abstract, self-contained system of each language, while la parole is the use of la langue to create actual utterances. Linguists, argued Saussure, should describe the properties of each langue without worrying about how it is instantiated in the minds and mouths of speakers. These are problems for the neighbouring sciences of psychology, physiology and physics. Sapir’s avowal of the formal autonomy of languages can be understood as part of this trend, even though at the same time he clearly did not wish to entirely relinquish his Humboldtian heritage, with its psychological and anthropological concerns.

There was a desire to break the spell of language, to revolt against its tyranny

But what about the linguistic determinism of the so-called Sapir-Whorf hypothesis? Although neither Sapir nor Whorf ever formulated any precise, testable proposition postulating the influence of language on thought, they certainly envisaged such effects. In 1929, Sapir wrote:

The fact of the matter is that the ‘real world’ is to a large extent unconsciously built up on the language habits of the group … The worlds in which different societies live are distinct worlds, not merely the same world with different labels attached … We see and hear and otherwise experience largely as we do because the language habits of our community predispose certain choices of interpretation.

Sapir and Whorf’s rhetoric answered to a contemporary moral panic about the use and abuse of language. The young 20th century saw public discourse perverted by new forms of propaganda, disseminated by such new technologies as radio and film, all of which accompanied and facilitated the catastrophic upheavals of the First World War and the political polarisation that resulted in the rise of totalitarian governments across Europe. There was a desire to break the spell of language, to revolt against its tyranny supporting irrationality and barbarity, and make it the servant of enlightened thought. This sentiment found expression in, among other places, the linguistic turn taken by the incipient analytic philosophy of this period. At the popularising end of the spectrum, innumerable manuals on meaning appeared, such as The Meaning of Meaning (1923) by C K Ogden and I A Richards, Science and Sanity (1933) by Alfred Korzybski, and The Tyranny of Words (1938) by Stuart Chase. This is the world of Orwell’s Newspeak, in which language is the master of mind.

Sapir and Whorf eagerly advertised the contribution their field of linguistics could make to solving these problems. In revealing the diversity of realities created by languages, linguistics could help to expose how language misleads us. In 1924, Sapir wrote:

Perhaps the best way to get behind our thought processes and to eliminate from them all the accidents or irrelevances due to their linguistic garb is to plunge into the study of exotic modes of expression. At any rate, I know of no better way to kill spurious ‘entities’.

By the mid-20th century, language-critical discourse had died down and academic linguistics largely returned to the dispassionate scientism familiar from the end of the previous century. Assessing several claimed cases of links between language structure and culture across a diverse range of languages, the US linguist Joseph Greenberg (1915-2001) declared mid-century: ‘[O]ne does not find any underlying semantic patterns such as would be required for the semantic system of a language to reflect some over-all world view of a metaphysical nature.’

Greenberg was inspired by the work of Boas and Sapir, and reignited the torch of linguistic typology that had been raised up by Gabelentz at the end of the 19th century. Greenberg’s continuation of the old Humboldtian project of investigating the structural diversity of the world’s languages while disavowing any connection between structure and cognition or culture was decisive for the later development of typological theory. What for Gabelentz had been the ‘highest task’ of language research was now officially off limits in the last corner of linguistics concerned with capturing and comparing the grammatical character of languages.

Interest in diversity was in any case at a low point in the mid-20th century. In his pursuit of ‘universal grammar’, Noam Chomsky (1928-) strove to re-establish a kind of psychic unity of mankind. The differences between individual languages, on Chomsky’s account, are mere phantoms, superficial variations on the same underlying system produced by an innate faculty of language shared by all humans. The linguist’s task should not be to meticulously catalogue these variants, but to factor them out and discover the universal principles governing all languages. Following Chomsky’s lead, received opinion in most quarters of academic research maintained this fastidious separation of language and thought until the end of the 20th century.

But linguistic relativity would not suffer banishment. Linguists and psychologists who could not ignore these questions have brought relativity back into the academic mainstream and delivered solid results. To name just one example, in ongoing, cutting-edge work, researchers have shown that certain languages may allow their speakers to unlock senses that are the common possession of all humans but remain unutilised by most people. In English and many other languages, spatial location is usually described in egocentric terms. If a fly were to land on my leg, I might say: ‘A fly has landed on the right side of my leg.’ Right is an egocentric spatial term that orients objects in the world according to an imaginary left-right axis projected from my body.

We are all, in a sense, compasses. English speakers are, mostly, not consciously aware of this

However, this is not the only way we can conceptualise space. In the Gurindji language, spoken in northern Australia – as in many other languages of the world – locations are usually described using the cardinal directions north, south, east and west. Assuming that I am sitting so my right leg is oriented towards the west, the equivalent sentence in Gurindji would be: ‘Karlarnimpalnginyi nyawama wurturrjima, walngin ngayinyja wurturrjila.’ Literally: ‘This is the outer upper west of (my) leg. The fly landed here on my leg.’ If I were to turn around and face the opposite direction, the fly would still – in egocentric terms – be on the right side of my leg, but a Gurindji speaker would point out that – in cardinal terms – the fly is now on the eastern part of my leg. While my private left-right axis might follow me around dutifully, the earth will always stand still.

The cardinal directions are not unknown in English, but they are typically employed only when talking on a geographical scale. By contrast, in Gurindji, even parts of the speaker’s body are located in a world-spanning co-ordinate system. Most English speakers would be at a loss to even identify the cardinal directions without the aid of a compass. How do Gurindji speakers do it? It would seem that they draw on a number of environmental cues, chief among these the course of the sun through the sky. But human neurophysiology is also sensitive to the magnetic field of Earth: the human brain responds in measurable ways to ambient magnetic fields. We are all, in a sense, compasses. English speakers are for the most part not consciously aware of this, even though their brain activity changes when surrounding magnetic fields are manipulated under experimental conditions. Recent experiments by the Australian linguist Felicity Meakins and her collaborators have shown that some Gurindji speakers can reliably report on shifts in ambient magnetic fields.

Gurindji speakers’ habit of using cardinal directions would seem to have opened up their powers of perception. At least some Gurindji speakers may be able to consciously feel Earth’s magnetic field. But do English speakers and Gurindji speakers live in ‘distinct worlds’, as Sapir would have it? Having greater sensitivity to some features of the environment still seems like something less than the all-encompassing, incommensurable worldviews of the Humboldtian tradition.

This is perhaps the chief source of the continuing scepticism regarding linguistic relativity in many academic quarters. We start with a feeling, an ineffable je ne sais quoi, that our language shapes our world. But to assess the truth of this claim, the scientist wants a hypothesis – a rigorous, experimentally testable statement of precisely how language shapes our world. Quasi-mystical meditations on my life in language are not the stuff of modern scientific journals. But any properly formulated hypothesis will necessarily be reductive and deflationary – devising empirical tests of the supposed differences in our worldviews inevitably means transforming our innermost feelings into detached, foreign objects that we can observe and analyse from the outside. Such tests can arguably never capture the totality and primordiality of the original feeling.

Does this mean that the scholarship of previous centuries has no place in today’s world or, alternatively, that modern science simply cannot fathom the philosophical depths explored by earlier work? Past and present scholarship are complementary. The writings of earlier scholars – however speculative they may seem to us now, and whatever problematic assumptions they may be built upon – undeniably capture something of our human experience and can inform the investigations of present-day researchers. In turn, the hypotheses and experiments of latter-day linguists and psychologists provide another perspective – shaped by the scientistic worldview of our own era – on these enduring questions of the connections between mind and language. In all these cases, we cannot even make sense of the questions without understanding something of the specific intellectual contexts in which they have arisen.