How Language Shapes Thought: The Cognitive Architecture of Words

In 1954, anthropologist Benjamin Lee Whorf published an essay describing how Hopi language lacks tenses for past, present, and future, instead marking whether events are manifested or not-yet-manifested. He argued this fundamentally altered Hopi time perception compared to English speakers' linear past-present-future model. Can linguistic structure shape how we perceive reality itself?

The question sparked decades of debate. Strong versions of linguistic determinism—language determines thought—have been largely rejected. But modern cognitive science reveals something more subtle and profound: language doesn't imprison thought, but it influences what we notice, remember, categorize, and reason about. Words are cognitive tools that make certain patterns of thinking easier and others harder.

Consider: Russian speakers distinguish between light blue (goluboy) and dark blue (siniy) as fundamentally different colors, not shades. This linguistic distinction makes them faster at discriminating these colors perceptually. English speakers learning Mandarin develop better understanding of temporal metaphors (viewing time as vertical, not just horizontal). Professional linguists debating whether language shapes thought often overlook the irony—they're using language to think about language shaping thought.

This analysis examines how language influences cognition: the mechanisms, the evidence, the limits, and the practical implications. We'll explore perception, categorization, memory, reasoning, and how multilingualism reveals language's role in shaping the architecture of mind.

---

The Sapir-Whorf Hypothesis: History and Versions

Origins: Edward Sapir and Benjamin Lee Whorf

Edward Sapir (1884-1939): American linguist and anthropologist. Studied Native American languages, noting structural differences from European languages.

Benjamin Lee Whorf (1897-1941): Chemical engineer turned linguist. Studied Hopi, Mayan languages. Argued linguistic structure shapes worldview.

Historical context: Early 20th century anthropology revealed profound diversity in language structure. This challenged universalist assumptions that thought patterns are innate and identical across humans.

Strong Version: Linguistic Determinism

Claim: Language determines thought. Speakers of different languages inhabit different cognitive realities. Some thoughts are impossible without appropriate linguistic structures.

Whorf's examples:

• Hopi time: Lacking tenses, Hopi speakers supposedly think about time differently—not as linear flow but as manifested vs. unmanifested events.
• Eskimo snow: (Often misattributed to Whorf) Claim that Eskimo languages have dozens of words for snow, enabling fine discriminations English speakers cannot make.

Status: Largely rejected. Empirical testing revealed:

• Hopi does have temporal markers, just structured differently than English.
• Many "Eskimo snow words" were inflections (like English has "snowflake," "snowball," "snowstorm").
• Thoughts can occur without specific words—speakers describe concepts their language lacks using circumlocution or borrowing.

Why it failed: Confuses linguistic encoding (how language packages information) with conceptual capacity (what minds can represent). Lacking a word doesn't prevent the concept.

Weak Version: Linguistic Relativity

Claim: Language influences thought—makes certain patterns of categorization, attention, and reasoning more habitual or accessible, without strictly determining possibilities.

Status: Supported by substantial evidence. Modern research shows linguistic structure affects:

• Perception: What we notice and discriminate.
• Categorization: How we group experiences.
• Memory: What we encode and recall.
• Reasoning: Spatial, temporal, and causal thinking patterns.
• Attention: What aspects of experience are salient.

This is the scientifically defensible position: Language doesn't limit thought but shapes its habitual patterns. Different languages provide different cognitive toolkits—tools influence what's easy, natural, efficient, though not strictly possible.

---

Language and Perception: Color, Space, and Objects

Color Perception: The Classic Case

Question: Do languages with different color terms cause speakers to perceive colors differently?

Background: Languages divide color spectrum differently:

• English: Basic terms for red, orange, yellow, green, blue, purple, pink, brown, black, white, gray.
• Russian: Distinguishes siniy (dark blue) and goluboy (light blue) as separate basic colors, not shades.
• Himba (Namibia): Five basic color terms grouping colors differently than English—what English calls "green" spans multiple Himba categories.

Evidence:

1. Discrimination speed: Russian speakers discriminate between siniy and goluboy faster than English speakers distinguish light and dark blues. The linguistic boundary facilitates perceptual boundary (Winawer et al., 2007).

2. Categorical perception: Colors within the same linguistic category seem more similar than equally-spaced colors across category boundaries. English speakers see blue-green boundary sharply; languages lacking this boundary show less perceptual distinctiveness.

3. Visual search: Searching for target color among distractors is faster when target crosses linguistic boundary (e.g., finding blue among greens for English speakers) (Gilbert et al., 2006).

4. Interference: The effect disappears or weakens when:

• Right hemisphere only: Color naming is left-hemisphere function. When right hemisphere tested alone (preventing language access), categorical perception weakens.
• Verbal interference: Concurrent verbal task (remembering number sequence) reduces color categorization effects, suggesting language actively involved during perception (Roberson & Davidoff, 2000).

Interpretation: Language doesn't change retinal input but influences online processing—attention, labeling, memory encoding. Color perception involves interaction between sensory input and linguistic categories.

Limits: Non-linguistic animals and pre-linguistic infants show some categorical color perception (universal perceptual tendencies), but language enhances boundaries and makes distinctions more automatic.

Space and Navigation: Frames of Reference

Question: Do languages describing space differently influence spatial reasoning?

Types of spatial reference:

• Egocentric (English default): "The cup is to my left." Reference frame centered on speaker's body. Rotates with speaker.
• Absolute (Tzeltal, Guugu Yimithirr): "The cup is north of me." Reference frame fixed to cardinal directions. Independent of speaker orientation.

Example: In English, describing object locations uses left/right/front/back relative to viewer. In Tzeltal (Mayan language, Mexico) and Guugu Yimithirr (Australia), descriptions use absolute directions—"The cup is north of the plate" rather than "left of the plate."

Cognitive effects (Levinson, 2003):

1. Memory encoding: When remembering spatial arrangements and recalling in different orientation:

• English speakers remember egocentric relations (cup was on left) and get confused when body orientation changes.
• Guugu Yimithirr speakers remember absolute relations (cup was north) and maintain accuracy regardless of body orientation.

2. Gesture: When describing past events or absent scenes:

• English speakers gesture relative to current body orientation (if describing leftward motion, gesture leftward regardless of actual direction).
• Tzeltal speakers gesture in absolute direction (if motion was eastward, gesture east regardless of how they're facing).

3. Mental maps: Navigational strategies differ—egocentric-language speakers better at route-based navigation ("turn left, then right"), absolute-language speakers better at survey-based navigation (maintaining orientation in large-scale space).

Interpretation: Linguistic habits train default spatial reasoning. English speakers can use absolute directions but don't habitually. Tzeltal speakers must track cardinal directions constantly because their language requires it.

This creates cognitive difference: Spatial attention, memory, and reasoning adapt to linguistic demands. The brain builds tools needed for linguistic task, and those tools become default for non-linguistic spatial thinking.

Object Categorization: Substance vs. Count

English distinction: Count nouns (countable discrete objects—"three apples") vs. mass nouns (continuous substances—"water," not "three waters").

Japanese: Lacks this grammatical distinction. Nouns aren't inherently count or mass—quantification uses classifiers regardless of substance/object status.

Cognitive effect (Imai & Gentner, 1997): Shown novel objects and asked whether similar items are "the same kind":

• English-speaking children focus on shape for solid objects (count nouns) and material for non-solid substances (mass nouns).
• Japanese-speaking children show less shape-bias, more flexible categorization.

Interpretation: Grammatical count/mass distinction trains attention to shape vs. material. This influences ontological assumptions—what defines object identity.

Time Perception: Metaphors and Thinking

Metaphors for time:

• English: Time as horizontal motion. "Looking forward to the future." "Putting past behind us." "Days ahead."
• Mandarin: Also uses vertical metaphors. "Next month" is xià gè yuè ("down month"). "Last month" is shàng gè yuè ("up month").
• Aymara (Andes): Past is ahead (visible, known), future behind (unseen, unknown)—reversing common metaphor.

Cognitive effects:

1. Spatial priming: When Mandarin speakers think about time after vertical spatial priming (e.g., seeing vertical arrangement), they process temporal sequences faster. English speakers don't show this (Boroditsky, 2001).

2. Gesture: English speakers gesture forward for future; Aymara speakers gesture forward for past (Núñez & Sweetser, 2006).

3. Reasoning: Mandarin-English bilinguals reason about time differently depending on language context—priming Mandarin activates vertical spatial thinking about time; priming English activates horizontal thinking.

Interpretation: Linguistic metaphors structure abstract domains (time) through concrete domains (space). This influences reasoning, memory, and attention in temporal thinking.

---

Language and Memory: Encoding, Retention, and Recall

Eyewitness Memory: The Loftus Effect

Elizabeth Loftus (1970s): Demonstrated language profoundly affects memory.

Classic study: Participants watched car accident video. Later asked either:

• "How fast were the cars going when they hit each other?"
• "How fast were the cars going when they smashed each other?"

Results: "Smashed" produced higher speed estimates and false memories—more participants later "remembered" broken glass (none existed) when asked with "smashed" (Loftus & Palmer, 1974).

Mechanism: Linguistic framing at encoding influences what's stored. Words activate schemas (expectations, associations) that blend with perceptual memory. Later retrieval can't distinguish actual perception from linguistically-induced inference.

Implications:

• Legal: Interrogation phrasing affects testimony reliability.
• Clinical: Therapeutic language can create false memories (memory recovery controversy).
• Media: News framing shapes public memory of events.

Memory for Events: Linguistic Encoding Advantage

General principle: Events easier to remember when language provides clear encoding.

Example: Remembering shades of color. English speakers better remember whether seen color was "blue" or "green" than specific shade within category—linguistic label provides retrieval cue (Lupyan, 2012).

Cross-linguistic differences: Russian speakers remember subtle blue distinctions (siniy vs. goluboy) better than English speakers because they encode linguistically at finer granularity.

Mechanism: Language provides conceptual hooks for memory. Non-linguistic memory (images, feelings) is possible but often fuzzier and harder to retrieve. Verbal labels anchor memories.

Caveat: Not all memory is linguistic—procedural memory (riding bike), emotional memory, spatial memory often operate without language. But for complex events, language facilitates encoding and organization.

Narrative Memory: Life Stories

Autobiographical memory shaped by narrative language:

1. Structure: Life events remembered as stories with coherent narrative arc—beginning, middle, end, causation, themes. Languages with different narrative conventions (e.g., emphasizing collective vs. individual agency) shape memory organization.

2. Self-concept: Identity emerges from narratives we tell about ourselves. Language provides categories (shy, ambitious, victim, survivor) that become self-schemas organizing memory.

3. Cultural memory: Shared historical events remembered through linguistic framing. "Liberation" vs. "invasion," "terrorist" vs. "freedom fighter"—language shapes collective memory.

---

Language and Reasoning: Logic, Causation, and Agency

Counterfactual Thinking: Grammar of Hypotheticals

Counterfactuals: Reasoning about "what if" scenarios contrary to fact.

Linguistic marking:

• English: Subjunctive mood signals counterfactual. "If I were rich, I would travel." (Distinguished from indicative: "If I am rich, I will travel.")
• Chinese: No grammatical counterfactual marking—context signals hypothetical status.

Cognitive effect (Bloom, 1981, contested by later research): Initial studies suggested Chinese speakers found counterfactual reasoning harder due to lack of grammatical marking. Subsequent research showed effect was artifact—Chinese speakers reason counterfactually equally well; linguistic form differs but capacity doesn't.

Interpretation: Grammatical marking makes counterfactuals explicit, but doesn't determine reasoning ability. Languages without overt marking use context and pragmatics. Reasoning capacity is universal; linguistic scaffolding varies.

Causation and Agency: Who's Responsible?

Linguistic differences:

• English: Agentive transitive. "John broke the vase." Subject is agent causing action.
• Spanish: Can use reflexive for accidental actions. "Se rompió el jarrón" ("The vase broke itself") removes agency from subject.

Cognitive and social effects (Fausey & Boroditsky, 2011):

1. Memory for agents: English speakers better remember who caused accidents. Spanish speakers remember what happened but less reliably remember agent when event framed non-agentively.

2. Blame attribution: Language primed with agentive framing increases blame. Non-agentive framing reduces blame and punishment severity.

3. Eyewitness testimony: Linguistic framing affects legal testimony—"He broke the window" vs. "The window broke" biases responsibility attribution.

Implication: Grammar encodes default assumptions about intentionality and agency, influencing moral cognition—judgments of responsibility, blame, punishment.

Numerical Cognition: Counting Systems and Math

Counting systems:

• Regular (Asian languages—Chinese, Japanese, Korean): Systematic structure. Eleven is "ten-one," twenty-four is "two-ten-four."
• Irregular (English, German): Opaque structure. Eleven, twelve (irregular); thirteen (three-ten structure, but backward).

Cognitive effects:

1. Early math learning: Children learning regular number languages grasp place-value concepts earlier and perform basic arithmetic faster (Miller et al., 1995).

2. Transparency: Regular systems make mathematical structure linguistically explicit. Saying "two-ten-four" embeds place-value meaning; saying "twenty-four" obscures it.

3. Long-term effects: Advantage persists into higher math—languages with regular number structure correlate with better PISA math scores (controlling for education spending and cultural factors).

Interpretation: Linguistic transparency reduces cognitive load for mathematical concepts. Easier encoding frees working memory for computation.

---

Bilingualism and Cognitive Flexibility

Thinking in Two Languages

Do bilinguals think differently in each language?

Evidence: Yes, systematically:

1. Emotion: Bilinguals report different emotional intensity in each language, especially for taboo words or childhood language (emotional memories encoded in native language feel more visceral) (Pavlenko, 2012).

2. Personality: Bilinguals self-report different personalities in each language—more assertive in one language, more polite in another. Reflects cultural norms embedded in language use (Ramírez-Esparza et al., 2006).

3. Moral judgments: Trolley problem presented in second language increases utilitarian responses (sacrificing one to save five)—emotional distance from second language reduces affective influence on moral reasoning (Costa et al., 2014).

4. Risk perception: Financial decisions become more risk-tolerant when problems framed in second language—"foreign language effect" reduces loss aversion (Keysar et al., 2012).

Cognitive Benefits of Bilingualism

Executive function: Bilinguals show advantages in:

• Inhibitory control: Suppressing irrelevant information (e.g., Stroop task).
• Task switching: Shifting between mental sets.
• Working memory: Maintaining and manipulating information.

Mechanism: Constantly managing two languages—suppressing one while using other, switching contexts—trains executive control.

Debate: Some recent studies fail to replicate executive function advantages, suggesting effects may be smaller or more context-dependent than initially thought. Publication bias may have exaggerated early findings.

Regardless: Bilingualism expands conceptual repertoire—provides alternative ways to categorize experience, frame problems, and think about domains. This is cognitive enrichment, even if executive function benefits are debated.

---

Can You Think Without Language?

Non-Linguistic Thought: Evidence

1. Visual thinking: Mental imagery, spatial rotation, visual pattern recognition occur without language. Architects, designers, visual artists report thinking in images.

2. Procedural knowledge: Skilled motor actions (playing piano, riding bike, throwing ball) involve procedural memory and sensorimotor systems, not linguistic representation.

3. Emotional intuition: Gut feelings, affective responses, implicit preferences often pre-linguistic—we feel before articulating reasons.

4. Animal cognition: Non-linguistic creatures (chimps, dolphins, crows) exhibit problem-solving, planning, tool use, social reasoning. Thought exists without language.

5. Aphasics: Patients with severe language impairment (Broca's or Wernicke's aphasia) retain many cognitive abilities—recognize faces, navigate environments, solve spatial puzzles, understand social situations.

6. Pre-linguistic infants: 6-12 month olds show understanding of object permanence, causation, numerosity, physical reasoning (Spelke, 2000)—core cognitive capacities pre-date language.

Where Language Is Critical

Complex abstract reasoning: Language especially important for:

1. Explicit symbolic reasoning: Logic, mathematics, scientific theory—rely heavily on linguistic and symbolic representation. Difficult to imagine propositional logic without language-like syntax.

2. Hierarchical compositional thought: Recursive embedding ("She thinks that he believes that they know...") leverages syntactic structure. Non-linguistic representations struggle with nested propositions.

3. Reflective consciousness: Introspection often takes form of internal monologue. Language enables thinking about thinking—metacognition.

4. Cultural knowledge: Much abstract knowledge (history, philosophy, science) transmitted and represented linguistically. Learning advanced domains requires language.

5. Hypothetical reasoning: Imagining non-actual scenarios, planning distant futures, moral reasoning about abstract principles—scaffolded by linguistic representation.

Linguistic Relativity in Non-Linguistic Thought?

Question: If language shapes thought, does it shape non-linguistic thought (e.g., visual imagery)?

Mixed evidence:

• Spatial reasoning in Tzeltal speakers uses absolute directions even in non-verbal tasks (arranging physical objects, mental rotation).
• Color perception effects require verbal interference to eliminate, suggesting active linguistic involvement during perception.
• But procedural and affective thought seem relatively insulated from linguistic influence.

Interpretation: Language pervades cognition more than initially assumed—influences "non-linguistic" domains through online processing (actively labeling, categorizing during perception and reasoning). But purely sensorimotor and implicit cognition operate independently.

---

Practical Implications: Using Language to Think Better

1. Vocabulary Expands Thought

Principle: Learning new words provides new conceptual tools.

Example:

• Sonder (Dictionary of Obscure Sorrows): Realizing each passerby has life as vivid and complex as yours. Having word makes concept more salient and retrievable.
• Schadenfreude (German): Pleasure from others' misfortune. English lacks single word, making concept slightly less accessible cognitively.
• Ikigai (Japanese): Reason for being, combining passion, mission, vocation, profession. Encapsulates complex concept efficiently.

Application: Deliberately expanding vocabulary—especially for nuanced psychological, social, or emotional concepts—increases cognitive precision and expressive capacity.

2. Frame Awareness

Principle: Language frames shape perception and decision-making. Awareness enables meta-framing.

Examples:

• Medical: "90% survival rate" vs. "10% mortality rate"—same fact, different emotional impact (Tversky & Kahneman, 1981).
• Political: "Estate tax" vs. "death tax"—frames affect support.
• Personal: "I failed" vs. "I learned"—reframing influences motivation and self-concept.

Application: When facing important decisions or communications, deliberately generate alternative framings. How would this sound in different words? What assumptions does my language embed?

3. Multilingual Thinking

Principle: Bilinguals can leverage different languages for different cognitive tasks.

Applications:

• Emotional distance: Making difficult decisions in second language reduces affective bias (foreign language effect).
• Creative perspective: Switching languages sometimes unlocks new problem framings or associations.
• Cultural concepts: Some languages offer concepts unavailable or less developed in others.

Practice: Even monolinguals can benefit from learning language fragments—exposure to alternate categorizations and framings enriches cognitive toolkit.

4. Precision and Clarity

Principle: Vague language enables vague thinking. Precision forces conceptual clarity.

Example: Replacing "interesting" with specific evaluation (surprising, elegant, puzzling, unsettling) clarifies why something is interesting.

Application: Writing and speaking with precision (avoiding jargon, cliché, vagueness) improves thought clarity. If you can't articulate clearly, often you haven't understood clearly.

5. Metaphor Awareness

Principle: Abstract reasoning relies on metaphors. Recognizing metaphors reveals hidden assumptions.

Examples:

• "Idea marketplace": Implies competition, winners/losers, supply-demand. Alternative: "idea ecosystem"—implies diversity, niches, symbiosis.
• "Time is money": Implies scarcity, efficiency focus. Alternative: "time is garden"—implies cultivation, growth, patience.

Application: Notice metaphors structuring your thinking. Experiment with alternative metaphors to reveal different dimensions of problems.

---

Conclusion: Language as Cognitive Tool and Scaffold

Core insight: Language doesn't determine what's thinkable, but it powerfully influences what we habitually notice, how we categorize experience, what we remember, and how we reason.

Mechanisms:

1. Perceptual influence: Linguistic categories shape attention and discrimination during perception.
2. Memory encoding: Words provide labels anchoring memories, making linguistically-encoded experiences easier to recall.
3. Reasoning scaffolds: Language structures abstract reasoning—temporal, spatial, causal, moral—through metaphors, grammar, and conceptual categories.
4. Cultural transmission: Language carries conceptual schemes, values, and knowledge across individuals and generations.

Variability and universality: Languages differ profoundly in structure and conceptual organization, creating measurable cognitive differences. But all languages enable complex thought—differences are in habitual patterns, efficiency, and cultural embedding, not fundamental capacity.

Practical implications: Understanding language-thought relationship enables:

• Cognitive enhancement: Expanding vocabulary, learning languages, using precise language improves thinking.
• Frame awareness: Recognizing linguistic framing enables conscious reframing and reduced manipulation.
• Communication effectiveness: Choosing words strategically influences others' perceptions, memories, and decisions.

Future research: Neuroscience, computational linguistics, and cross-cultural cognitive science continue revealing how language and thought interact—through brain imaging, corpus analysis, and behavioral experiments across diverse languages.

Final reflection: The very debate about whether language shapes thought illustrates the point—we think with language about language shaping thought. Language is both the object of analysis and the tool enabling analysis. This reflexivity reveals language's deep integration with cognition. We don't just think about language; we think through language, making words not mere labels but cognitive technologies extending and structuring mind itself.

Word count: 4,846 words

---

References

1. Bloom, A. H. (1981). The Linguistic Shaping of Thought: A Study in the Impact of Language on Thinking in China and the West. Lawrence Erlbaum Associates.
2. Boroditsky, L. (2001). Does language shape thought?: Mandarin and English speakers' conceptions of time. Cognitive Psychology, 43(1), 1-22. DOI: 10.1006/cogp.2001.0748
3. Costa, A., Foucart, A., Hayakawa, S., et al. (2014). Your morals depend on language. PLOS ONE, 9(4), e94842. DOI: 10.1371/journal.pone.0094842
4. Fausey, C. M., & Boroditsky, L. (2011). Who dunnit? Cross-linguistic differences in eye-witness memory. Psychonomic Bulletin & Review, 18(1), 150-157. DOI: 10.3758/s13423-010-0021-5
5. Gilbert, A. L., Regier, T., Kay, P., & Ivry, R. B. (2006). Whorf hypothesis is supported in the right visual field but not the left. PNAS, 103(2), 489-494. DOI: 10.1073/pnas.0509868103
6. Imai, M., & Gentner, D. (1997). A cross-linguistic study of early word meaning: Universal ontology and linguistic influence. Cognition, 62(2), 169-200. DOI: 10.1016/S0010-0277(96)00784-6
7. Keysar, B., Hayakawa, S. L., & An, S. G. (2012). The foreign-language effect: Thinking in a foreign tongue reduces decision biases. Psychological Science, 23(6), 661-668. DOI: 10.1177/0956797611432178
8. Levinson, S. C. (2003). Space in Language and Cognition: Explorations in Cognitive Diversity. Cambridge University Press. DOI: 10.1017/CBO9780511613609
9. Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13(5), 585-589. DOI: 10.1016/S0022-5371(74)80011-3
10. Lupyan, G. (2012). Linguistically modulated perception and cognition: The label-feedback hypothesis. Frontiers in Psychology, 3, 54. DOI: 10.3389/fpsyg.2012.00054
11. Miller, K. F., Smith, C. M., Zhu, J., & Zhang, H. (1995). Preschool origins of cross-national differences in mathematical competence: The role of number-naming systems. Psychological Science, 6(1), 56-60. DOI: 10.1111/j.1467-9280.1995.tb00305.x
12. Núñez, R. E., & Sweetser, E. (2006). With the future behind them: Convergent evidence from Aymara language and gesture in the crosslinguistic comparison of spatial construals of time. Cognitive Science, 30(3), 401-450. DOI: 10.1207/s15516709cog0000_62
13. Pavlenko, A. (2012). Affective processing in bilingual speakers: Disembodied cognition? International Journal of Psychology, 47(6), 405-428. DOI: 10.1080/00207594.2012.743665
14. Ramírez-Esparza, N., Gosling, S. D., Benet-Martínez, V., Potter, J. P., & Pennebaker, J. W. (2006). Do bilinguals have two personalities? A special case of cultural frame switching. Journal of Research in Personality, 40(2), 99-120. DOI: 10.1016/j.jrp.2004.09.001
15. Roberson, D., & Davidoff, J. (2000). The categorical perception of colors and facial expressions: The effect of verbal interference. Memory & Cognition, 28(6), 977-986. DOI: 10.3758/BF03209345
16. Spelke, E. S. (2000). Core knowledge. American Psychologist, 55(11), 1233-1243. DOI: 10.1037/0003-066X.55.11.1233
17. Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453-458. DOI: 10.1126/science.7455683
18. Winawer, J., Witthoft, N., Frank, M. C., Wu, L., Wade, A. R., & Boroditsky, L. (2007). Russian blues reveal effects of language on color discrimination. PNAS, 104(19), 7780-7785. DOI: 10.1073/pnas.0701644104