Steven Pinker: Linguistics as a Window to Understanding the Brain | Big Think

Main ideas, concepts, and lessons

Language as a “window” into the mind

Steven Pinker frames himself as a cognitive scientist (not a linguist by training) and argues that language should be studied not just as an object, but as evidence about how the human brain and mind work.

Language is central because it:

• distinguishes humans from other species,
• enables cooperation and shared knowledge through words,
• raises major scientific questions (evolution of language; how the brain computes it).

The “miracle” and expressive power of language

Humans can decode meaningful information from continuous sound streams (breath-exhaled noise) that encode information. With small changes in those sounds, humans can express virtually any topic, from entertainment to deep scientific questions.

This expressive power motivates the scientific study of language.

Universality and cultural breadth

• Language exists in every human society studied and is complex in all cases.
• The video notes about 6,000 languages worldwide, emphasizing that no studied human culture lacks complex language.
• It references the Tower of Babel story and Darwin’s observation that children show an instinctive tendency to speak more than they show instincts for other cultural skills.

A structured view of what “language science” studies

Language sciences include study of:

• Grammar (how words combine into phrases/sentences),
• Phonology (sound patterns),
• Semantics (meaning),
• Pragmatics (how meaning works in conversation).

Related subfields:

• Psycholinguistics: real-time processing,
• Language acquisition: how children learn,
• Neurolinguistics: how language is computed in the brain.

---

What language is not (important distinctions)

Not writing

• Spoken language exists across human cultures; writing is much rarer and historically recent.
• Alphabetic writing is said to have appeared only once (attributed to the Canaanites).
• Children typically do not have an instinct to write; it must be learned.

Not prescriptive “proper grammar”

Linguists distinguish:

• Descriptive grammar: how people naturally speak,
• Prescriptive grammar: rules of “careful” writing.

The video argues that prescriptive rules often don’t make sense (e.g., split infinitives, double negatives) and are tied to historical artifacts (e.g., Latin analogies or dialect-specific usage).

Not thought itself

The video argues that language is not identical to thought:

1. Nonverbal cognition exists: babies/animals can understand cause-effect, objects, intentions without speech.
2. Visual thinking happens without language: people solve spatial problems by mental imagery (e.g., mentally rotating shapes) rather than word-by-word description.
3. Memory retains gist, not exact wording: people can remember meaning but not reproduce exact sentences.
4. Understanding requires non-linguistic processing: even simple phrases (example: “Wet hair, lather, rinse, repeat.”) require tacit knowledge to interpret correctly.
5. If language were only thought, acquisition becomes puzzling: language grows “grassroots” rather than being designed fully by an external intelligence.

Linguistic relativity is acknowledged but bounded

• The Sapir–Whorf / linguistic relativity idea is mentioned: language can affect thought.
• But the video stresses that language is not the same as thought and not all mental life is sentence recitation.

---

How language works (main framework)

The video divides language into three core topics plus interfaces.

1) Words (lexicon / mental dictionary)

Words are arbitrary signs

• Cites Ferdinand de Saussure: the link between word form and meaning is arbitrary.
• Example: “duck” does not resemble a duck physically, but users have memorized the mapping between sound and concept.

The mental lexicon is huge

Using dictionary-sampling style estimates:

• typical high school vocabulary: ~60,000 words
• roughly one new word every two hours from about age 1.

---

2) Grammar (rules / algorithms for building sentences)

Influenced by Noam Chomsky.

Grammar must explain creativity/productivity

Speakers can produce and understand new sentences they have never heard. Therefore, language knowledge is not merely memorized sentence lists; it’s an internal algorithm/recipe.

Syntax is not the same as meaning

Example:

• “Colorless, green ideas sleep furiously.” It’s near-meaningless but still recognized as syntactically well-formed.

Syntax cannot be reduced to simple word-to-word probabilities

Many adjacent word pair transitions are near-zero, yet the full sentence is interpretable.

Long-distance dependencies

Words far apart constrain each other:

• “either …” typically requires “or”
• “if …” typically expects “then”

Embedded/nested dependencies occur in complex structures.

Hierarchical/phrase-structure building

Sentences are described as hierarchical trees (“upside down tree”):

• sentence → noun phrase (subject) + verb phrase (predicate)
• verb phrases expand into verb + noun phrase (object) + complement, etc.

Why phrase-structure rules matter

• Open-ended creativity: explains unlimited generation.
• Expression of unfamiliar meanings: new combinations convey new “news.”
• Combinatorial explosion: the number of possible expressions is described as infinite in principle.

Illustration with “world’s longest sentence”

• Mentions Guinness’s claim (1,300 words from Faulkner).
• The argument: it can’t be “the longest” because additional embeddings (e.g., “Guinness noted that…”) can always extend it.

Phrase structure supports ambiguity resolution

Example:

• “On tonight’s program, Conan will discuss sex with Dr. Ruth.” Different parses yield different meanings about what is being discussed with whom.

Language acquisition and rule learning by children

Children can’t memorize all sentences; they extract rules. Evidence includes:

• two-word stage: children produce combinations not heard verbatim (e.g., “more outside”),
• overgeneralization of grammar rules (past tense errors like “holded,” “sticked”).

Wug test

• Teach: “This is a wug.”
• Show: “Now there are two of them.”
• Children produce “wugs,” demonstrating productive rule use.

Universal grammar claim Chomsky’s argument: poverty of the input—children nevertheless learn complex structure-dependent rules.

• Example: forming English questions shows children choose structured transformations rather than simplistic word-by-word movement.

Critiques of Chomsky

• Universal grammar may not be specific to language (could reflect general mind abilities).
• Not proven across all ~6,000 languages.
• Alternative general learning models (e.g., neural networks) might explain acquisition without special grammar-specific knowledge.

---

3) Phonology (sound patterns)

Phonology defines possible “words”

It defines which sequences count as possible “words” for a language. Some sound sequences feel imaginable as English words, yet are impossible under English phonological formation rules.

Sound adjustment and allomorphy

Example: past tense “-ed” has different pronunciations (“walked,” “jogged,” “patted”) depending on surrounding sounds.

Foreign accent via transfer

L2 learners carry phonological rules from their first language, producing accent.

Deliberate manipulation

When people intentionally manipulate phonology for effect, it becomes poetry and rhetoric.

---

Language interfaces: input and output between language and the brain

Production interface (brain → speech sounds)

How vocal anatomy produces speech:

• the larynx/voice source vibrates to create rich harmonic patterns,
• the vocal tract shapes/amplifies/suppresses harmonics,
• tongue/lip positioning changes resonant cavities, producing distinct categories of vowels/consonants.

Evolutionary compromise (Darwin)

• The larynx’s descent improves speech ability but increases choking risk (maladaptive anatomy).
• The Heimlich maneuver is mentioned as a solution.

Comprehension interface (world sounds → brain)

Speech recognition is hard for computers but easy for humans, due to:

1. Co-articulation: the same phoneme can sound different depending on neighboring sounds (e.g., similar “c” sounds in “Cape Cod” can be produced differently).
2. Lack of clear segmentation: the speech waveform doesn’t contain obvious pauses marking word boundaries.
   • Humans use internal lexical knowledge to identify word boundaries.
   • Computers lacking that knowledge struggle.

Word-boundary-related wordplay

• Songs/jingles exploit how the continuous sound stream can be parsed into different word sequences (nonsense vs grammatical sentences).

Context and ambiguity

• Humans effortlessly use context and common sense to determine intended meanings.
• Computers can produce comically wrong translations (legend example):
  • “The spirit is willing, but the flesh is weak” → “The vodka is agreeable, but the meat is rotten.”

Pragmatics interface (meaning in context)

Central principle: Cooperative Principle Assume the conversation partner is trying to be truthful, clear, and cooperative.

Example:

• “If you could pass the guacamole, that would be awesome.” Interpreted as a polite request, not a literal hypothetical.

Humor example:

• Robots ignoring pragmatic meanings (e.g., “give me a hand” interpreted literally).

Pronoun resolution Understanding “he” referring to an antecedent not explicitly named requires:

• background knowledge about people/relationships,
• conversational inference, which is difficult to program for computers.

---

Practical applications mentioned

• Computer speech understanding and generation
• Diagnosis/treatment of language disorders
• Teaching reading, writing, and foreign languages
• Interpreting language in law, politics, and literature

---

Overall conclusion (lesson)

Language is described as a distinctively human natural wonder:

• It uses a finite set of mental tools (lexicon + grammar) to express an effectively unlimited number of ideas.
• It must be studied through its connections to cognition, biology, evolution, society, and the mind.
• Language is framed as a window into human nature, not just a cultural artifact like writing or prescriptive grammar.

---

Speakers / sources featured (as stated or quoted)

• Steven Pinker (speaker; Professor of Psychology at Harvard)
• Charles Darwin (quoted/paraphrased about human speech instinct)
• Ferdinand de Saussure (source for “arbitrariness of the sign”)
• Noam Chomsky (major influence on modern grammar; universal grammar; poverty of the input)
• Sapir–Whorf (linguistic relativity / Sapir–Whorf hypothesis)
• Julius Caesar (mentioned in the context of split infinitives analogy)
• William Faulkner (associated with the “world’s longest sentence” example)
• Guinness Book of World Records (cited for the longest-sentence claim)
• Captain Kirk (from Star Trek, used as an example in split infinitives discussion)
• J. T. (Mick) Jagger / The Rolling Stones lyric (double-negative example: “I can’t get no satisfaction”)
• Heimlich Maneuver (named as a solution to choking risk; attributed to invention by implication)
• Darwin (again: evolutionary commentary on larynx descent)
• Bartlett’s Familiar Quotations (referenced as where Chomsky’s quote is said to appear)
• TV Guide (source of the ambiguous “Conan will discuss sex…” sentence example)
• Mairzy Doats (example of word-boundary wordplay)
• “Fuzzy Wuzzy” (example sentence)
• Doggerel / the rhyme starting “I scream, you scream…” (example of wordplay/pronunciation sequence)