S1-301 – Without words, no world

How language makes thought possible, and what it makes impossible

In the 1970s, a team of researchers at Stanford University studied a group of deaf children in Nicaragua who had been brought together for the first time in a newly established school for the deaf. The children had no shared language: each had developed their own idiosyncratic system of gestures at home, understood by their families but not by anyone else. When they were placed together, something remarkable happened. The children began spontaneously combining their gesture systems, and within a generation, a new language had emerged: Nicaraguan Sign Language. What made the emergence remarkable was not merely that the children had invented a language (that, in itself, was extraordinary enough) but that the language they invented had grammatical structure. It had ways of marking who did what to whom, of indicating whether an event was completed or ongoing, of embedding one proposition inside another. The children had not been taught these structures. They had created them, apparently from the demands of the communicative project itself. Language, the episode seemed to suggest, is not merely a tool that thought uses to express itself. It is something that thought requires to become fully itself.¹

This is a more radical claim than it initially sounds. The everyday model of the relationship between language and thought is one of expression: we have a thought, and then we find words for it. The thought is primary; the language is a vehicle. What the Nicaraguan Sign Language case (and a body of research in cognitive linguistics that has accumulated over the past half century) suggests is that this model is, in significant ways, backwards. Language is not merely the vehicle of thought. It is, for much of human thought, the road that makes certain destinations reachable at all.

This article examines what language does to thought: not merely how it expresses thought but how it constitutes it, limits it, and makes possible kinds of thinking that would otherwise be unavailable. It is the first of four articles on language and communication in this series, and it establishes the framework within which the subsequent articles operate.

The Vygotsky thesis: language as internalized social speech

The psychologist Lev Vygotsky argued in the 1930s, in work that was suppressed in the Soviet Union and did not reach Western readers until decades later, that language does not begin as an internal tool for thought and subsequently become external in speech. It begins as external (as social speech between people) and only gradually becomes internalized as the silent inner speech through which mature human thought conducts itself.² The young child does not think in words and then speak. They speak, and through speaking and hearing themselves speak and being responded to by others, they gradually internalize the social exchange and transform it into private thought.

This developmental sequence has direct implications for the relationship between language and thought. If thought is internalized social speech, then the structure of thought is not independent of the structure of language: it is shaped by it, organized through it, and in significant part constituted by it. The concepts available to a thinker are the concepts that the language of their community provides. The distinctions the thinker can draw are the distinctions the language marks. The categories through which experience is organized are the categories that the language has named.

This does not mean that thought is impossible without language. Pre-linguistic infants think: they have expectations, they are surprised by violations, they track objects through occlusion, they respond to the intentions of others. Animals think: they solve problems, they recognize individuals, they make choices, they use tools. What Vygotsky’s thesis suggests is that the specifically human form of complex, abstract, culturally transmitted thought (the thought that can range freely across time, that can entertain counterfactuals, that can reason about reasoning itself) depends on the internalization of linguistic structures that are themselves the product of a social history far longer than any individual life.

The practical implication is that what a person can think is not independent of what language their community speaks and what concepts their language provides. This is not the strong Sapir-Whorf hypothesis (the claim, now largely discredited, that language determines thought completely and that speakers of different languages cannot share the same thoughts). It is the weaker and more defensible claim that language influences thought: that having a word for something makes it easier to think about, to remember, to discuss, and to build upon, and that the absence of a word for something makes it correspondingly harder to work with as a concept.

The vocabulary gap and what it costs

The practical demonstration of the influence of vocabulary on thought is available in every domain where specialized vocabulary does real cognitive work. The experienced physician who has the word “proprioception” (the sense of the body’s own position and movement in space) can think about a class of neurological symptoms that the physician who lacks the word must describe in roundabout terms each time and therefore cannot manipulate as efficiently in reasoning. The legal professional who has the concept of “estoppel” (the principle that a party is prevented from asserting a position inconsistent with one that they have previously adopted) can apply that concept rapidly to new cases, transfer it across contexts, and combine it with other legal concepts in ways that require the concept to be available as a unit. The philosopher who has the concept of the modus tollens (the inference form that if P implies Q, and Q is false, then P is false) can apply it quickly and recognize its violations without having to reconstruct the logical structure each time.

In each case, the word is not merely a label for a thought that existed without it. The word is the cognitive tool that makes the thought rapidly and reliably available: that allows it to be stored, retrieved, transferred, and combined with other thoughts with an efficiency that would be impossible if the thought had to be reconstructed from scratch each time. This is the chunking process from article S1-202, applied to concepts: the word is the chunk that provides access to the associated structure, and the richness of the associated structure depends on the conceptual work that the word’s community of users has done with it over time.

The vocabulary gap (the differential access to concepts that is produced by differential exposure to domains of knowledge) is therefore not merely a gap in information. It is a gap in cognitive tools. The child who grows up in an environment that provides rich conceptual vocabulary in a domain (whose parents talk about history, whose books use scientific concepts, whose community engages with political ideas) has access to thinking tools in those domains that the child without that exposure lacks. The gap is not primarily a gap in what the children know. It is a gap in what they can efficiently think.³

This observation has implications that go beyond the developmental and into the political. Democratic deliberation (the process by which citizens form views on complex collective questions and communicate those views through the political process) requires the ability to think with concepts that are not always intuitively available. The concept of an externality (a cost or benefit that falls on parties not involved in the transaction that produces it) is not intuitively obvious. Neither is the concept of a confidence interval, or a selection effect, or a counterfactual. These are not ornamental intellectual luxuries. They are the tools without which certain kinds of reasoning about policy are impossible. A citizenry without access to them is not merely less informed; it is less equipped to think about the domains that democratic governance requires thinking about.

What concepts without names look like, and what they cost

The complement of the vocabulary question is the question of what happens to concepts for which there is no word. Can we think something we have no word for?

The answer, as the research suggests, is: partially, slowly, and with great difficulty. Pre-linguistic children and languageless animals demonstrate that some forms of thought are available without language: they can categorize, anticipate, remember, and choose without words. But the thought that is available without language is thought that can be conducted through perception, action, and embodied engagement with the immediate environment. The thought that ranges freely across time, that entertains abstract relationships between things not present, that considers multiple possibilities simultaneously and reasons about their implications (this thought, the specifically human thought that has produced science, philosophy, law, and literature) requires language as its medium.

The German language is often cited for its capacity to name concepts that other languages lack words for: Schadenfreude (pleasure at another’s misfortune) and Weltschmerz (the pain of the world, the sadness that comes from comparing the world as it is to the world as it should be) are familiar examples. Less familiar but equally instructive is the Japanese concept of ma (the meaningful pause, the pregnant silence, the space between things that is itself a thing) which is very difficult to think about in languages that have no equivalent, not because the experience is unavailable to non-Japanese speakers but because holding it as a concept, reasoning about its implications, and communicating it efficiently requires a word that those languages do not provide.

The absence of a word does not make the experience unavailable. It makes the concept (the experience packaged as a manipulable unit of thought) much harder to use. The English speaker who has an experience of ma may not recognize it as such, may not notice it recurring in different contexts, may not be able to communicate it efficiently to others, and may not build on it conceptually in the way that a concept with a name can be built on. The experience exists. The concept, in the full cognitive sense (the thing that can be stored, retrieved, combined, transferred, and built upon) is much less available.

McGilchrist and the two languages

Iain McGilchrist’s account of the divided brain, discussed in article S1-201 and developed throughout the series, provides a further dimension to the relationship between language and thought that is worth introducing here.⁴ McGilchrist’s argument is that the two hemispheres of the brain engage with the world in fundamentally different ways: the left hemisphere through analysis, categorization, and the manipulation of explicit representations, and the right hemisphere through contextual, embodied, and holistic engagement that resists explicit representation.

Language, in McGilchrist’s account, is primarily a left-hemisphere tool. The left hemisphere’s engagement with language is analytical and categorical: it breaks experience into nameable components, applies labels, manipulates propositions, and constructs arguments. This is the language that the preceding sections of this article have been examining: the language that constitutes concepts and enables abstract reasoning.

But McGilchrist argues that there is a form of engagement with the world that the right hemisphere conducts and that language, in the left-hemisphere sense, cannot fully capture. The right hemisphere’s understanding is implicit, contextual, and sensitive to tone, nuance, and the unspoken (it understands what is meant by a piece of music, by a facial expression, by a metaphor, by the feel of a situation) in ways that the left hemisphere’s explicit representation cannot fully encode. The right hemisphere, in this account, knows more than it can tell, in exactly the sense that Polanyi described with tacit knowledge.

The implications for language as a tool of thought are double-edged. Language, in the left-hemisphere sense, is indispensable for the kinds of thought that require explicit representation, logical manipulation, and precise communication. But it is also, in McGilchrist’s account, a potential trap: the tendency to treat what can be said as all there is to say, to reduce the richness of experience to what the naming apparatus can capture, to mistake the label for the thing it names in exactly the way that Feynman’s father was warning against. The Conscious Look, in this dimension, is the practice of maintaining awareness that what the left hemisphere’s language can say is not the complete account of the world: that the right hemisphere’s understanding of context, nuance, and implicit meaning is a form of knowledge that the naming apparatus cannot fully encode.

The social construction of thought, and its limits

The observation that thought depends on language, and that language is a social product (developed by communities over generations, carrying the distinctions and categories that those communities have found useful) implies that individual thought is not as individual as it appears. The concepts through which a person thinks, the distinctions they can draw, the categories they can use, are the concepts, distinctions, and categories that their linguistic community has developed and transmitted. The thinker is not the author of their conceptual tools. They are the inheritor and user of tools that were built by others and that carry, in their structure, the assumptions and priorities of the communities that built them.

This does not make thought merely conventional or the individual merely a passive recipient of inherited categories. The history of science, philosophy, and art is the history of individuals and communities extending, revising, and sometimes replacing the conceptual tools they inherited: introducing new distinctions, naming previously unnamed phenomena, challenging categories that had been treated as natural and revealing them as constructed. Language is both the medium within which thought must work and the product that thought can modify.

But it does mean that the categories through which we think are not neutral. They carry assumptions. They foreground some distinctions and background others. They make some things easy to say and others difficult or impossible. The concept of race, as currently used in political discourse, foregrounds certain physical characteristics as the relevant dimension of human categorization and backgrounds the enormous within-group variation that makes the category biologically meaningless while making it socially consequential. The concept of the economy, discussed in article S1-203, makes it easy to think about aggregate economic activity and harder to think about the specific distributional consequences that aggregate measures conceal. The concept of a terrorist makes it easy to think about the individual actor and harder to think about the structural conditions that produce the action.

None of this means that the concepts are wrong; they may capture something real. It means that they are not innocent. They select, they emphasize, they foreground, they background. The practice of The Conscious Look, applied to language, is the practice of asking, about the concepts through which one is currently thinking: what does this concept make easy to see, and what does it make hard? Whose distinctions are these, and what was the purpose for which they were developed? What would the situation look like if it were described with different concepts? The questions are not designed to produce universal skepticism about language; thought requires categories, and categories require choices. They are designed to maintain awareness that the categories are choices, and that different choices would make different aspects of the world visible.

Notes

¹ The emergence of Nicaraguan Sign Language has been studied most extensively by the linguist Judy Kegl and her colleagues, and by Ann Senghas at Barnard College. The findings have been reported in several papers, most accessibly in Senghas, A., Kita, S., and Özyürek, A. (2004). Children creating core properties of language: Evidence from an emerging sign language in Nicaragua. Science, 305(5691), 1779-1782. The episode is particularly instructive because it shows the emergence of grammatical structure in a population that had no exposure to any signed or spoken language with that structure, providing the closest available natural experiment to the conditions under which language might emerge from scratch. The grammatical regularities that appeared in Nicaraguan Sign Language (particularly the use of spatial grammar to indicate relationships between actors and objects) were not present in the home gesture systems of the founding children and were not taught by the hearing educators at the school, who were attempting to teach lip-reading and spoken Spanish.

² Vygotsky, L. S. (1934/1962). Thought and Language. MIT Press. Vygotsky’s work was suppressed in the Soviet Union shortly after his death in 1934, partly because of its incompatibility with Pavlovian behaviorism, which was the officially approved framework for psychology under Stalin. It was not translated into English until 1962, and not widely available in the West until the 1970s and 1980s. The developmental sequence Vygotsky describes (from social speech to private speech to inner speech) has been confirmed by subsequent developmental research and is now one of the best-established findings in developmental psychology.

³ The connection between vocabulary and cognitive capacity has been developed most systematically in the educational research literature. Hart, B., and Risley, T. R. (1995). Meaningful Differences in the Everyday Experience of Young American Children. Paul H. Brookes Publishing. The Hart and Risley study, which tracked language use in homes across socioeconomic levels, found a gap of approximately 30 million words heard by age three between children from high-income and low-income households, a gap that correlated strongly with subsequent vocabulary development, academic performance, and cognitive outcomes. The finding has been replicated, debated, and refined in subsequent research, but the basic observation (that differential exposure to rich vocabulary in early childhood produces differential cognitive tools) is robust.

⁴ McGilchrist, I. (2009). The Master and His Emissary: The Divided Brain and the Making of the Western World. Yale University Press. McGilchrist’s treatment of language as a primarily left-hemisphere tool is one of the more controversial aspects of his thesis, and it should be held with appropriate tentativeness: the neuroscience of language lateralization is more complex and more bilateral than older accounts suggested. The more defensible version of his claim is that the analytical, categorical, and propositional aspects of language are more strongly lateralized to the left hemisphere, while the pragmatic, contextual, and metaphorical aspects of language are more dependent on right-hemisphere processing. This weaker version is consistent with the current neuroscientific evidence and is sufficient to support the argument made in this article.