Imagine enjoying a picnic as a friend suddenly informed you that there was a spider on your north shoulder. Would you know which shoulder to check? This might sound bizarre to an uninitiated English speaker, but to a speaker of Gugu Yimithirr, it would be perfectly natural. Speakers of this Australian language (and several others around the world) primarily use absolute directions when describing everyday situations, to the exclusion of relative directions like “left” and “right”. Similarly for speakers of Tzetal (a language spoken in mountainous southern Mexico), everyday spatial relationships are described in terms of “uphill” and “downhill,” reflecting the terrain where the language is spoken. Speakers of languages that favour absolute direction are reported to have highly accurate of awareness of absolute directions without the aid of a compass, even indoors or in the absence of obvious landmarks.
This is a striking example of how differences in language may affect our perception of the world around us. The idea that language and thought are intertwined is not a new one. Plato suggested that “The soul when thinking appears to me to be just talking.” In more modern times, Benjamin Lee Whorf described it as “the world is presented in a kaleidoscopic flux of impressions which has to be organized by our minds–and this means largely by the linguistic systems in our minds.” This idea that language shapes how we think and see the world goes by several names, including the Sapir-Whorf Hypothesis, Linguistic Relativity, and the Whorfian Hypothesis. But how true is it really?
Traditionally, views have been described as either “universalist” or “relativist,” but these views can encompass a wide range of different ideas. The strongest forms of the relativist view include the notion that language constrains the range of thoughts available to us (often called linguistic determinism) or even that we do all of our thinking through language.
The idea that we think in a language has some intuitive appeal. However, it’s pretty easy to show that this can’t be entirely true. Steven Pinker, a prominent critic of linguistic determinism, pointed out that “Sometimes it is not easy to find any words that properly convey a thought. When we hear or read, we usually remember the gist, not the exact words, so there has to be such a thing as a gist that is not the same as a bunch of words.” Our ability to interpret ambiguous sentences also hints at thought that exists independently of language. When we read (or hear) a headline like the famously ambiguous “British Left Waffles on Falkland Islands” (often attributed to the Guardian), we can apply two different meanings—one involving indecision of a political party during the 1982 Falklands War and the other involving an abandoned breakfast treat. It could be argued that these multiple interpretations (two or more “thoughts” for one string of words) wouldn’t be possible if language and thought were one and the same. If people couldn’t think beyond the confines of their language, it would also be impossible to create new words—a process that we know occurs regularly–just look at the number of words we invent every year for things on the internet.
On the other hand, we do know that language can facilitate memory and helps us grasp complex concepts like numbers, directions, and understanding the viewpoints of others (also known as Theory of Mind). People who grow up with limited early language exposure (often due to hearing loss and lack of access to sign language), experience difficulties and delays in a variety of cognitive domains. In this sense, language does appear to affect how we think. What is less clear is whether (or how much) learning different languages results in different patterns of thinking.
Debate between relativist and universalist sides has often been heated and recent research acknowledges that the strongest forms of both views are unlikely to be true. However, before discussing evidence, it’s worth noting that this field of research remains controversial. The validity and significance of nearly all experimental findings on both sides of the debate have been questioned in some form or another.
But let’s dive into it shall we?
At a very basic level, language experience almost certainly affects how we perceive certain sounds. For example, children of English and Hindi speakers respond differently to the sounds /t̪/ (dental t, pronounced with the tongue touching the front teeth and /ʈ/ (retroflex t, pronounced with the tongue further back than English /t/). To English speaking adults, these sounds can be difficult to distinguish, both sounding like some sort of /t/. However, for Hindi speakers, these sounds are clearly different, as different as, say, /t/ and /d/ in English. Interestingly, infants are born with the ability to tell these sounds apart. At six months, babies with English-speaking parents can distinguish these sounds easily, but by the age of one year, English-learning infants lose the ability to distinguish them, while Hindi-learning infants keep it. Similar patterns are followed for numerous sounds in languages from around the world. These differences show how language can change our perception at a very basic level, but this is not what people usually think of when they think about Linguistic Relativity. What about other domains of perception not directly linked to the mechanics of language itself?
To begin with here, let’s discuss time.
Most, if not all, languages use spatial metaphors to describe time, but differ in how they achieve this. English tends to describe time as moving horizontally, with the future in front of us and the past behind. We talk about looking forward to something or putting the past behind us. Mandarin Chinese, on the other hand, often uses vertical metaphors, with the past above and the future below (although it does use horizontal metaphors too). The Mandarin word for next is the same as the word for down and the word for previous is the same as the word for up. This leads to the interesting question of whether speakers of these two languages actually think of time differently. And if that’s true, how do we measure it? Fortunately, psycholinguists are here to do this so that you don’t have to.
A few different methods have been used to investigate these questions. In one experiment, participants were asked to arrange pictures, such as photographs of a person at several different ages, sequentially. English speakers almost exclusively arranged the pictures horizontally from left to right, but Mandarin speakers arranged them vertically (from top to bottom) fairly often, around 30% of the time. In similar tasks, speakers of Hebrew (which is written from right to left) arranged pictures from right to left, opposite to the direction English speakers favored. A completely different pattern was found in speakers of Aboriginal Australian languages that use absolute direction (north, south, east, and west, rather than left and right). They tended to arrange the pictures from east to west, mirroring the direction of the sun’s daily journey across the sky. Experiments like this show an intriguing relationship between language and time perception, but it’s not always easy to determine whether language actually causes the differences. What if the language differences exist because of cultural differences in how time is conceptualized? This kind of chicken-and-egg problem can be difficult to avoid when studying the relationship between language and culture.
One potential way to get around this issue is to study bilingual speakers and observe whether they perform differently in their two languages. Experiments with Mandarin-English bilinguals have shown that they are more likely to give responses indicating a vertical perception of time when tested in Mandarin than when tested in English. In another experiment, Mandarin speakers were more likely to indicate vertical thinking about time when answering questions asked with vertical metaphors. These kinds of within-language results (which have also been found in other domains) show that language may affect how we think in the moment, but also that thinking can be quite flexible—not necessarily defined just by what language we speak.
Moving on to the fascinating case of language and Color. Color perception has a long history of being studied in relation to cross-linguistic perceptual differences. At a physical level, modern humans all have essentially the same hardware for color vision (color blindness and rare genetic variants aside). We generally have three kinds of cone cells in our retina, each attuned to different wavelengths of light. This means that the colors we are physically capable of seeing don’t vary across populations. However, the way language is used to describe color is highly variable across languages. The case of color demonstrates both striking differences in perception across language groups and evidence of universal patterns in cognition. This is still an active area of research and exactly how universal or variable color perception is remains controversial.
Many scholars have noted that Ancient Greek didn’t describe colors in the same way as modern English—classical texts by Homer have famously likened the appearance of the sea to wine or described sheep with a color that could also be applied to blood. When discussing the rainbow, Greek poet and philosopher Xenophanes, describes it as having only three bands of color. In an omission surprising to many English speakers, it appears that Ancient Greek didn’t have any word corresponding closely to English “blue” (although blue did exist within the scope of words that also covered other colors including greens and greys). In fact, this absence of a specific word for blue has been noted in numerous other languages around the world, especially those spoken in ancient times. Writing of ancient Vedic Hymns, philologist Lazarus Geiger observed that
“These hymns, consisting of more than 10,000 lines, are nearly all filled with descriptions of the sky. Scarcely any other subject is more frequently mentioned; the variety of hues which the sun and dawn daily display in it, day and night, clouds and lightnings, the atmosphere and the ether, all these are with inexhaustible abundance exhibited to us again and again in all their magnificence; only the fact that the sky is blue could never have been gathered from these poems by any one who did not already know it himself.”
English speakers may find this baffling, but in a worldwide linguistic context, lack of a specific category for blue isn’t particularly unusual. It’s so common for languages to have a single term that encompasses both green and blue that linguists even have a name for it– “grue.”
English has 11 basic color words (red, orange, yellow, green, blue, purple, black, white, pink, grey, and brown), but this set of colors is far from universal. Languages are usually described as having anywhere from two to (approximately) twelve basic colors. A survey of 20 languages in the 1960s found that color words in languages followed a surprisingly predictable pattern. Languages with only two basic color terms divided colors into “black” and “white,” with darker colors falling under the “black” category and lighter colors falling under the “white” category. In languages with three basic color terms, the third was always “red.” Fourth and fifth color terms were generally either “green” or “yellow.” Only with six color words was “blue” included, followed by “brown” at seven. Languages with higher numbers of color terms included some combination of “purple”, “pink”, “orange”, and “grey”. A few languages make a further division, such as Russian having two separate words for lighter blues and darker blues (much like the difference between red and pink in English). More recent data from larger and more diverse samples of languages (The World Color Survey included over 100) show that the patterns above largely hold true, although there are some exceptions and variability, including substantial variability between speakers of the same language.
With all this variety in how languages describe color, it’s hard not to wonder if language affects how we actually see color. Does the sky still look blue if you don’t have a word for it? Would a rainbow have different numbers of stripes depending on the language you speak? The answer (not to mention the question itself) is complicated. Understanding and measuring exactly how different cultures conceptualize color isn’t necessarily a straightforward task. Even defining what constitutes a color term in the first place can be tricky. It’s not always clear if words refer directly to color, or to other visual attributes, or similarity to some other object. Color terms can also be limited to specific domains (such as blonde primarily being used to describe hair). Some languages, such as Walpiri (spoken in central Australia), may not even have a word for the concept of “color” in the first place. All of that having been said, many researchers have attempted to untangle the relationship between language and color perception experimentally.
Several experiments have shown differences between speakers of different languages in tasks related to color perception and memory. For example, speakers of Russian (which has different basic color words for light and dark blue) are faster at identifying the difference between pairs of colors that straddle this boundary than pairs that don’t. When presented with the same stimuli, English speakers show no such differences. Similar results have been found in Himba (a language spoken in Namibia that doesn’t distinguish between green and blue) and Korean (which has a different boundary for green than English). Some recent work has suggested that these effects may be more pronounced in the right visual field, corresponding to the left hemisphere of the brain (where language is primarily processed). Additionally, these effects tend to subside when participants perform a verbal distraction task (such as mentally rehearsing an eight-digit number), suggesting that linguistic mental resources are required for the effects to occur.
On the other side of the debate, universalists have argued that small differences in highly specific experimental tasks don’t necessarily indicate any important difference in everyday thought and perception. We can still see differences between colors within a category, even if we don’t have different words for them. Steven Pinker pointed out that “No matter how influential language might be, it would seem preposterous to a physiologist that it could reach down into the retina and rewire the ganglion cells.” Also, while the number of color categories is highly variable from language to language, they tend to cluster around similar hues. Exact boundaries and “best examples” for specific color categories may differ across languages, but tend to be much more similar than would be expected by chance. Given the infinite number of ways it would be theoretically possible to divide up the color space, this consistency is remarkable. Even infants and children who can’t yet reliably label color categories show categorical perception of color that largely follows these patterns, hinting at some universal aspects of color perception.
This brings us to numbers. Number is another area where language differences can be striking. As with color, number shows a mixture of universal and relativist findings. While fairly unusual, there are languages that have limited to no words for expressing exact numbers. Pirahã, spoken by a small group of hunter-gatherers in the Amazon, is perhaps the most extreme case. It has been described as having no words for exact quantities at all—only approximate terms for “one,” “two,” and “many.” Mundurukú, also spoken in the Amazon, only has numbers up to five (and even “four” and “five” are not always exact). Perhaps unsurprisingly, speakers of these languages perform differently on mathematical tasks than people from cultures that regularly use large, exact numbers.
In a typical experiment, several nuts were placed in a can and then removed one at a time. After each nut was removed, participants were asked whether any nuts remained in the can. For quantities above about 3, Mundurukú listeners often didn’t respond accurately. However in estimation tasks (like comparing sets of dots), they perform much like speakers of other languages. Estimates tended to follow a logarithmic pattern in accordance with Weber’s Law (which predicts that differences of the same magnitude appear smaller for larger sets). That is, sets of 4 and 8 dots would be easier to distinguish than, say, 34 and 38 dots. Even for sets as large as 80 dots, Mundurukú speakers showed largely similar performance to French controls in some estimation tasks. Patterns of estimation consistent with Weber’s law have also been observed in children, infants, and even non-human animals, suggesting humans may have an innate estimation ability with ancient evolutionary origins.
Overall, findings seem to suggest universal patterns for quantity estimation, but number words may be necessary for tasks involving exact amounts above about three. However, the picture is not entirely clear. Even Mundurukú speakers who had learned some Portuguese showed similar patterns of results to monolinguals, so culture, rather than just language, may play a role in how we think about numbers. On the other hand, there is also research suggesting that both Pirahã and Mundurukú speakers who have received mathematical education (either through schooling or informal instruction) show higher accuracy in some quantitative tasks. It’s also worth remembering that some of these findings are controversial due to limited sample size and the difficulties of controlled experiments in remote settings.
Differences between languages extend further than what they have words for or how they label specific concepts. Can a language’s grammar affect how we think? Many languages assign grammatical gender to nouns, with everyday objects like tables, cups, and bridges being classified, often quite arbitrarily, as masculine or feminine. When asked to assign a voice to various inanimate objects, people tend to choose voices that match the grammatical gender of the noun. For example Spanish speakers would be more likely to assign a female voice to a house (la casa), which is feminine in Spanish and a male voice for a book (el libro), which is masculine in Spanish. However many other studies using less direct tasks measuring memory or associations have showed mixed results, so it’s not clear how pervasive the effect of grammatical gender on our thinking is.
One recent study suggests that grammar could also have an effect on memory. Languages are often classified as predominantly right branching (with the most important information at the beginning of a sentence or phrase) or left branching (with the most important information at the end). English primarily uses right branching structures (such as “turtles who live in the sewers”), but we do use some left-branching patterns as well (“sewer-dwelling turtles”). Both phrases describe a kind of turtle, but differ as to whether we find that out at the beginning or end of the phrase.
This study found that speakers of left-branching languages remembered early items in lists of words or numbers more accurately than late items while speakers of right-branching languages showed the opposite pattern and instead remembered late items more accurately. As with other areas of study, caution is warranted and results have to be interpreted carefully. Only eight languages were covered in this study and more research is needed to determine whether these findings generalize to a larger sample of languages. If these results do hold up to scrutiny over time, they suggest that language could have a subtle, but pervasive effect on basic thought processes like memory.
So to sum up, despite decades of extensive study, the exact details of how language may affect thought remain slippery and elusive. Language doesn’t appear to constrain how we think, but it does seem to be able to influence our thoughts more subtly. It may nudge us towards certain ways of thinking, make certain distinctions easier to notice, or facilitate our ability to grasp certain concepts. Given the diversity of the world’s languages (many estimates place the number around 7,000) and the difficulty of measuring thought, debate about the specifics of the relationship between language and thought is likely to continue for many decades to come.
If you liked this article, you might also enjoy our new popular podcast, The BrainFood Show (iTunes, Spotify, Google Play Music, Feed), as well as:
- If Children Grew up Isolated from Adults, Would they Create Their Own Language?
- What is the Record for Most Languages Spoken By One Person?
- Do Kids Really Learn Languages Faster Than Adults?
- How Far Back in Time Could a Modern English Speaker Go and Still Communicate Effectively?
Expand for References
Absolute Direction Gugu Yimithirr Tzeltal
Review Articles Linguistic Relativity Effect of Language on Perception Wikipedia
Pinker Quote “Sometimes, it is not easy…”
Early Language Exposure Risk of Language Deprivation
Infant sound perception Cross Language Speech Perception Becoming a Native Listener
Time How Languages Construct Time Absolute Direction and Time Do English and Mandarin Speakers Think About Time Differently English and Mandarin Time in 3D Immediate and Chronic Influence of Spatio-temporal metaphors
Color Language, Thought, and Color Color Naming Universals Focal Colors are Universal After All No Universals in Color Perception Russian Blue Perception Categorical Perception of Color in Korean Human Color Perception English and Himba Toddlers Color Perception in Infants Color Terms Ancient Greek Ancient Greek (continued) Xenophanes Lazarus Geiger Wine Dark Sea
Numbers Exact and Approximate Arithmetic Effect of Education Number as Cognitive Technology Quantity Recognition Among Speakers of an Anumeric Language Weber’s Law Independence of Language and Mathematical Reasoning
Gender Grammatical Gender and Linguistic Relativity Spanish Grammatical Gender
Memory Word Order Predicts Working Memory
Number of Languages in the World How Many Languages?