The Biological Foundations of Human Language and the Creative Aspect of Speech
Introduction: The Study of Language
We are at the outset of the work, where Chomsky establishes the fundamental orientation of his inquiry: language is not merely a cultural artifact but a specific biological endowment of the human species.
The opening passages present language as “a system of knowledge” that is generative in nature. Chomsky immediately distinguishes his approach from behavioral or structuralist accounts that treat language as a set of habits or a collection of utterances. He argues that any speaker of a language possesses a generative grammar — a finite system of rules that specifies the infinite set of well-formed sentences in that language. This system is “a representation of the speaker’s competence,” meaning the underlying knowledge that allows a person to produce and understand sentences they have never encountered before. The actual use of language in concrete situations, which Chomsky terms performance, is distinct from this underlying competence, though it is guided by it. Performance can be affected by memory limitations, distractions, errors, and other non-linguistic factors, but the study of language must focus on the system of knowledge itself.
Chomsky then introduces the central problem that his work will address: Plato’s problem, or the problem of how children acquire such rich and complex knowledge of language given the limited and often impoverished data they receive. A child hears only a finite, often imperfect, sample of sentences, yet by a certain age they command a mature grammar that generates an infinite range of novel expressions. The input (the primary linguistic data) is insufficient to determine the adult grammar solely through induction or general learning mechanisms. This gap between input and output, Chomsky argues, can only be explained if the child brings to the task a highly structured innate endowment — a Universal Grammar that constrains the possible forms of human language. This innate faculty is what makes language acquisition possible, and it is “a distinctive property of the human mind.”
What this means within the work is that Chomsky is reorienting linguistics away from the mere description of languages toward the study of the human mind itself. Language is being presented as a cognitive organ, a biological system with its own principles and structures, not merely a tool for communication. The entire subsequent investigation will be driven by this core assumption: that to understand language is to understand a specific component of human nature.
Chapter 1: Generative Grammar and the Notion of “Knowledge of Language”
Here we enter the first major technical exposition, where Chomsky lays out the formal structure of a generative grammar and explains what it means to say that a speaker “knows” a language.
Chomsky begins by specifying that a generative grammar is “a system of rules that in some explicit and well-defined way assigns structural descriptions to sentences.” This is not a trivial notion: the rules must be explicit, capable of generating every well-formed sentence without relying on the intuition of the speaker after the grammar is formulated. The structural descriptions assigned by the grammar include information about the phonetic form, the meaning, and the underlying syntactic organization of each sentence. A grammar, then, is a theory of the language in question, and the linguist’s task is to construct such a theory and test it against the judgments of native speakers.
The key term introduced here is competence versus performance. Competence is “the speaker-hearer’s knowledge of his language,” while performance is “the actual use of language in concrete situations.” Chomsky emphasizes that performance is not a direct reflection of competence; it is influenced by many extra-grammatical factors. Therefore, the linguist must attempt to abstract away from performance errors and focus on the underlying system of knowledge. He writes:
“Linguistic theory is concerned primarily with an ideal speaker-listener, in a completely homogeneous speech-community, who knows its language perfectly and is unaffected by such grammatically irrelevant conditions as memory limitations, distractions, shifts of attention and interest, and errors (random or characteristic) in applying his knowledge of the language in actual performance.”
Having established this idealization, Chomsky turns to the nature of the rules themselves. A generative grammar is composed of a syntactic component, a phonological component, and a semantic component. The syntactic component is the central, generative part: it specifies the abstract deep structures and the transformations that map these to surface structures. The phonological component converts surface structures into phonetic representations, while the semantic component interprets deep structures to yield meanings. This three-part architecture — deep structure → transformational rules → surface structure — constitutes what Chomsky calls the standard theory of transformational generative grammar.
What this section accomplishes is the precise definition of the object of study. Chomsky is not merely describing sentences; he is arguing that the linguist’s theory must be formal enough to be tested, and that this theory is at heart a theory of what a speaker unconsciously knows. The distinction between competence and performance allows the field to advance, but it also carries a deeper implication: that the mind contains a specific, structured system of knowledge that is not reducible to behavior.
Chapter 2: The Structure of a Generative Grammar
We proceed deeper into the technical architecture, where Chomsky specifies the internal components of the grammar and how they interact to generate sentences.
The chapter opens with a detailed account of the base component of the grammar. The base consists of two parts: a set of categorial rules (often of the form X → Y Z, known as phrase structure rules) and a lexicon, which is a list of lexical items each with their syntactic, semantic, and phonological features. The categorial rules generate deep structures, which are abstract syntactic representations of sentences, while the lexicon provides the specific words that fill the positions defined by these structures. For example, the rule S → NP VP generates a sentence as a noun phrase followed by a verb phrase, and the rule VP → V NP generates a verb phrase as a verb followed by a noun phrase. These rules combine to produce a phrase marker — a hierarchical tree diagram that shows the grouping of constituents.
Chomsky then introduces transformations, which are operations that map deep structures to surface structures. A transformation can delete, insert, or reorder constituents. The classic example is the passive transformation, which takes an active deep structure (e.g., *the man hit the ball*) and converts it into a passive surface structure (e.g., *the ball was hit by the man*). The deep structure captures the underlying thematic relations (who did what to whom), while the surface structure reflects the actual order of words in the sentence. Transformations, Chomsky stresses, are structure-dependent: they operate on phrase markers, not on linear strings of words. This is a crucial point: a transformation cannot simply move the third word to the end; it must refer to the syntactic categories (e.g., “move the NP that is the object of the verb to the subject position”).
The recoverability condition is introduced here: every transformation must be such that the deep structure can be recovered from the surface structure. In other words, the system cannot arbitrarily delete information; the meaning must be determinable from the final form. This condition places strong constraints on what transformations are possible, and it ties the syntactic system to the requirement that sentences have determinate interpretations.
What Chomsky is building here is a theory of language that is both generative and explanatory. The rules are not merely descriptive lists but must be so constrained that they predict which sentences are grammatical and which are not, and they must account for the systematic relationships among sentences (such as the active-passive relation). The structure-dependence of transformations, in particular, reveals a property of human language that is not logically necessary — a machine could in principle operate on linear order — but that is universally true of natural languages. This fact will later be used as evidence for the innate structure of the language faculty.
Chapter 3: Deep Structures and Surface Structures
This chapter consolidates and deepens the distinction between the two levels of syntactic representation, showing how they interact with the interpretive components of the grammar.
Chomsky begins by clarifying that deep structure is not the same as meaning, nor is surface structure the same as sound. Rather, deep structure is the syntactic level that feeds into the semantic component for interpretation, while surface structure feeds into the phonological component for pronunciation. However, there is a crucial asymmetry: only deep structure determines meaning, while surface structure determines phonetic form. This is the standard hypothesis of transformational grammar at this point in the work. Chomsky provides the example of a sentence like *John is easy to please*, which has a surface structure that appears to make *John* the subject, but its deep structure reveals that *John* is the object of *please* (roughly, *it is easy for someone to please John*). The meaning depends on the deep structure, not the surface arrangement.
Chomsky then examines a range of constructions to demonstrate the necessity of deep structure. Consider the sentence *I expected the doctor to examine John* versus *I persuaded the doctor to examine John*. In the first, *John* is the object of *examine* at the deepest level; in the second, *the doctor* is also the object of *persuade* at the deep level. The surface structures appear similar, but the underlying grammatical relations are different, and these differences explain why the sentences behave differently under transformations such as passivization. Only by positing distinct deep structures can the grammar correctly capture the speaker’s knowledge.
A major point of clarification is that deep structures themselves are generated by the base rules, and there is no further level “behind” them. The deep structure is the initial syntactic representation from which all derivation proceeds. Chomsky also addresses the notion of universal base: the idea that the base rules (the categorial rules and the lexicon) are part of Universal Grammar and are therefore invariant across languages in their essential form. While languages may differ in their surface transformations, the deep structures they can generate are severely constrained by the innate language faculty.
The significance of this chapter is that it establishes deep structure as the level at which grammatical relations (subject, object, etc.) are defined, and it shows that these relations are not directly given by surface order. This explains how speakers can understand sentences that have the same meaning but different word orders, and how they can recognize that sentences with similar surface forms may have different meanings. The deep/surface distinction is not an abstract formalism; it captures something real about the mental representation of language.
Chapter 4: Transformational Rules and Constraints
We now turn to a more detailed investigation of the transformational component, focusing on the constraints that limit the power of transformations and make the theory empirically testable.
Chomsky begins by noting that if transformations were completely free — able to move any element to any position — the grammar would generate an infinite number of ungrammatical sentences. Therefore, transformations must be subject to constraints that restrict their operation. One of the most important constraints introduced here is the A-over-A constraint: if a transformation applies to a category of type A, it cannot apply to a smaller category of the same type contained within it. For example, if a rule moves a noun phrase, it cannot move a noun phrase that is embedded inside a larger noun phrase. This constraint prevents the grammar from generating sentences like *the man who saw the boy came* being transformed into *the boy came who saw the man*, which is ungrammatical.
Chomsky then discusses cyclic application of transformations: transformations apply first to the most deeply embedded clause, then successively to higher clauses, each time completing their operation before moving outward. This cyclic principle explains why certain transformations cannot apply across clause boundaries. For instance, the rule that moves a question word to the front of a sentence must operate within the clause where the word originates; it cannot move a word from a subordinate clause to the front of the main clause unless it moves stepwise through each clause boundary. Chomsky formalizes this with the subjacency condition, which states that no transformation can move an element across more than one bounding node (typically, a clause or a noun phrase) in a single step.
A further constraint is the specified subject condition: a transformation cannot involve an element that is subject to a specified subject within its domain. This condition blocks certain operations across a subject noun phrase, explaining why *John seems to know the answer* is grammatical (no subject intervenes) but *John seems that he knows the answer* is grammatical only as a different construction. These constraints, Chomsky argues, are not learned features of individual languages; they are part of the fixed architecture of the language faculty. They help solve Plato’s problem by dramatically reducing the range of grammars a child must consider.
The practical consequence of this chapter is that the theory of transformations becomes not just a description of how sentences are related, but a theory of what is possible and impossible in human language. The constraints are proposed as universal principles of the human language faculty, and they serve as evidence that the mind is not a blank slate but a highly structured system. By discovering such constraints, linguists are uncovering the innate structure that makes language acquisition possible.
Chapter 5: Phonology — The Sound Systems of Language
We are now moving from syntax—the structure of sentences—to the domain of sound. Chomsky turns to phonology, the component of grammar that governs how the abstract structures generated by syntax are realized as actual sound sequences.
Chomsky begins by distinguishing the phonetic representation—the physical sound produced by a speaker—from the phonological representation, a more abstract mental construct. The phonological component of a grammar takes the surface structure produced by the transformational rules and maps it onto a phonetic representation. This mapping is governed by a set of phonological rules that operate in a specific order. For example, the phonological component must account for why the plural morpheme in English is pronounced , [±nasal], [±continuant]. The phonological rules do not operate on whole segments (like “the sound /p/”) but on these features. A rule might specify: “Change a segment to [+voiced] when it occurs between two [+voiced] segments.” This feature-based approach allows the grammar to capture generalizations that would otherwise be scattered across many separate rules. For instance, the process of vowel harmony in Turkish or Hungarian can be stated as a single rule that spreads the feature [±back] across all vowels in a word, rather than as a list of individual vowel changes.
The ordering of phonological rules is crucial. Chomsky provides the example of the English plural formation, where a rule of epenthesis (inserting a vowel) must apply *before* a rule of devoicing (making a consonant voiceless). The underlying plural morpheme is /z/. The epenthesis rule inserts a vowel /ɪ/ before the /z/ when the stem ends in a consonant that shares the features [+strident] and [+continuant] (as in *buzz-es*). The devoicing rule then changes /z/ to [s] after a voiceless consonant (as in *cat-s*). If the rules applied in the reverse order, the wrong results would occur. This ordering is not arbitrary; it is systematic and reflects the grammar’s internal logic. The phonological component, like the syntactic component, is a system of rules with a specific architecture, and understanding that architecture is part of understanding the knowledge of language that a speaker possesses.
Clarification: This chapter demonstrates that phonology is not a random list of sounds and rules but a highly structured system operating on abstract features and ordered rule applications. The concept of distinctive features is the phonological parallel to the syntactic categories and deep structures of earlier chapters. Both domains are built on a small set of abstract primitives combined through rule-governed processes. By showing that the sound system, like the syntax, is rule-governed and ordered, Chomsky reinforces the central claim that language is a system of rule-governed creativity at every level, from the organization of sounds to the construction of the most complex sentences.
Chapter 6: The Formal Nature of Language
Chomsky steps back from specific analyses to ask a more abstract question: What kind of mathematical object is a grammar? In this chapter, he formally defines the concept of generative grammar within the framework of mathematical linguistics. He introduces the Chomsky hierarchy, a classification of formal grammars based on the kinds of rules they use, ranging from the most restrictive to the most powerful.
The hierarchy has four levels. Type 0 grammars are unrestricted—their rules can rewrite any string of symbols as any other string. These are equivalent to Turing machines in computational power. Type 1 grammars are context-sensitive—their rules can only rewrite a symbol in the context of surrounding symbols. Type 2 grammars are context-free—their rules rewrite a single non-terminal symbol as a string of symbols, regardless of context. Type 3 grammars are regular (or finite-state)—their rules are of the form A → aB or A → a, where A and B are non-terminals and a is a terminal symbol. Chomsky demonstrates that natural languages cannot be described by a finite-state grammar (Type 3), nor by a context-free grammar (Type 2) alone. They require the power of a context-sensitive grammar (Type 1) at the very least, and the transformational component adds another layer of complexity.
“A grammar, then, is a finite set of rules, specified in a system of representation, that generates an infinite set of sentences. The sentences themselves are strings of symbols, and the grammar imposes a structural description on each.”
The crucial insight of this chapter is that the structure of the grammar itself is a subject of empirical investigation. By asking what kinds of rules are needed to generate the sentences of a natural language, Chomsky is asking what kind of computational system the human mind uses. If a particular type of grammar (e.g., a finite-state one) is too weak to generate languages like English, then the human language faculty must be structured to employ a more powerful type. The hierarchy provides a way to measure the complexity of a language’s grammar and, by extension, the complexity of the mental computations required to understand and produce it. The fact that natural languages fall into the middle of the hierarchy (Type 1) is not a mathematical accident; it reveals something about the design of the human mind.
Clarification: This chapter grounds the earlier descriptive work in a rigorous mathematical framework. The Chomsky hierarchy is a foundational tool in both linguistics and computer science. Chomsky’s point is that the rules of a generative grammar are not just convenient fictions; they are hypotheses about the nature of the computational processes that the brain carries out when we use language. By showing that natural languages require a grammar of a specific formal type, he is making a strong claim about the innate structure of the mind: it is not a blank slate capable of learning any possible rule system; it is constrained by a specific formal architecture that only allows certain types of rules. This is a key step in his argument that the study of language is a branch of cognitive psychology or theoretical biology.
Chapter 7: Linguistics and Cognitive Psychology
In this pivotal chapter, Chomsky makes explicit the broader implications of his work. He argues that linguistics is not simply a branch of humanities but a subfield of cognitive psychology. The object of study is not merely a collection of sentences or texts but the internal mental grammar—the tacit knowledge that a speaker-hearer possesses. This knowledge, Chomsky contends, is a mental organ or cognitive faculty that grows in the mind as the child matures, much like the visual system or the immune system.
Chomsky introduces the distinction between competence (the speaker’s tacit knowledge of the language) and performance (the actual use of language in concrete situations). Performance is subject to all sorts of errors, memory limitations, and distractions. The task of the linguist is to study competence—the underlying system that makes performance possible. Therefore, the grammar a linguist writes is a model of competence; it is a theory of the knowledge that resides in the mind of the speaker. This is why Chomsky insists on the term generative grammar—it is not a manual for how to speak but a hypothesis about the mental system that enables a person to produce and understand an infinite number of sentences.
“The grammar of a language is a theory of the knowledge of the language, and the linguist’s task is to make this knowledge explicit.”
The implications are profound. If language is a mental organ, then its fundamental properties are not learned from the environment but are innate. The universal grammar is the genetic blueprint for this organ, specifying the principles that all languages must obey. The environmental input (the linguistic data a child hears) merely triggers the growth of the organ and sets a few parameters (like whether the verb comes before or after the object). This is Chomsky’s solution to Plato’s problem: how can children know so much about language given the impoverished data they receive? The answer is that they do not learn language from scratch; they grow it, guided by an innate language faculty that contains a rich set of constraints.
Clarification: This chapter is the bridge between the technical linguistics of the previous chapters and the philosophical arguments about innate knowledge that dominate the second half of the book. Chomsky redefines linguistics as a part of theoretical psychology, with the goal of discovering the universal principles of the human mind. The competence/performance distinction is crucial: it allows the linguist to abstract away from messy real-world data and focus on the ideal speaker-hearer. The claim that grammar is a mental representation—a piece of internal structure—is what makes the entire enterprise scientific. By studying language, Chomsky argues, we are studying the structure of the mind itself, and the findings of linguistics have direct bearing on the central questions of philosophy and cognitive science.
Chapter 8: Principles and Parameters — The Theory of Universal Grammar
We are now in the second major theoretical development of the work, where Chomsky moves beyond the model of a grammar as a system of specific rules for each language and proposes a radically different conception of Universal Grammar based on principles and parameters.
The presentation begins by re-examining the fundamental problem that motivated generative grammar from the start: how children acquire language so rapidly and uniformly despite limited and often degenerate input. Chomsky argues that the earlier model of generative grammar, with its language-particular rules (such as the specific transformational rules for English passives or interrogatives), actually made the acquisition problem *harder*, not easier. If children had to learn dozens of specific rules for each construction in their language, and if these rules varied arbitrarily across languages, then the poverty of the stimulus argument would be even more pressing. The child would need to choose not only which rules exist but also their precise formulation.
The solution, Chomsky explains, is to reconceive the system. Instead of a grammar being a list of rules, it should be seen as a system of principles that hold universally across all human languages, combined with a small set of parameters that allow for limited, binary variation. He writes: “The theory of Universal Grammar must specify a set of principles that hold for all languages, and a set of parameters that define the ways in which languages can differ.” The child’s task, then, is not to learn an entire grammar from scratch, but to set the parameters of the universal system based on the input they hear. For example, a principle might state that every sentence must have a subject, but a parameter might determine whether that subject must be overtly pronounced (as in English) or can be silent (as in Italian or Spanish — the pro-drop parameter).
This shift represents a profound change in what it means to “know a language.” Under the principles-and-parameters model, knowledge of language is not knowledge of a large set of idiosyncratic rules, but rather the possession of a fixed, innate system of principles with a few open parameters. Learning a particular language involves fixing these parameters to specific values. Chomsky clarifies the implications: “A language is not, then, a system of rules in the traditional sense; rather, it is a set of specifications for parameters within a fixed system of principles of Universal Grammar.” The grammar of English is simply the particular way in which the universal principles are realized when the parameters are set one way; the grammar of Japanese is the realization when they are set another way.
The importance of this move within the work cannot be overstated. It transforms the theory of Universal Grammar from a list of abstract properties that all languages share into a genuine explanatory mechanism for language acquisition. The poverty of the stimulus argument is now given a precise answer: children do not need to “learn” grammar because they already possess the universal principles; they only need to set the parameters by matching the input to the available options. This eliminates the problem of having to explain how children choose among infinitely many possible grammars. The competence of a native speaker is now seen as a specific parameter setting of a universal initial state, and the study of comparative syntax becomes the study of how different languages cluster their properties around different parameter choices.
Chapter 9: Logical Form and the Meaning of Sentences
Having established the formal structure of syntax and the principles-and-parameters framework, Chomsky now turns to the interface between syntax and meaning — the level of Logical Form (LF) . This chapter extends the Y-model of grammar first introduced in the discussion of deep and surface structures, where the derivation splits at surface structure: one branch goes to Phonetic Form (PF) (the sound side), and the other goes to Logical Form (the meaning side).
Chomsky begins by distinguishing between what a sentence *means* and what a speaker *intends* or what a sentence *refers to* in the world. He argues that Logical Form is a purely syntactic level — it is the representation of a sentence that is relevant to its semantic interpretation, but it is not the same as full semantics or pragmatics. LF encodes properties such as quantifier scope, anaphora, and binding relations. For example, the sentence “Everyone loves someone” is ambiguous: it can mean that there is one person everyone loves (wide scope for “someone”) or that for each person, there is some (possibly different) person they love (wide scope for “everyone”). This ambiguity, Chomsky argues, is not a matter of different surface structures but of different LF representations, derived by a special class of transformational operations that apply after surface structure.
A central operation at LF is Quantifier Raising (QR) , which moves quantifier phrases to positions where their relative scope can be read off the structure. Chomsky presents this as a transformational rule that is invisible to phonology — it applies after the sentence has been pronounced, so we do not hear the moved elements. He writes: “The rules that map S-structure to LF are of the same general nature as those that map D-structure to S-structure, but they apply to a different level and have distinct effects on interpretation.” This creates a picture where the syntactic derivation continues past the point of pronunciation, and the ultimate meaning of a sentence depends on processes that are hidden from the surface form.
The chapter also explores binding theory, the set of principles that govern the relationship between pronouns, reflexives, and their antecedents. Chomsky introduces the three binding principles: (A) an anaphor (like “himself”) must be bound in a local domain; (B) a pronoun (like “him”) must be *free* in a local domain; and (C) a lexical name (like “John”) must be free everywhere. These principles operate on LF representations, explaining why sentences like “John likes himself” are grammatical while “John likes him” (when “him” refers to John) is not. The entire system is presented as part of the innate Universal Grammar — children do not learn these constraints; they are built into the architecture of the language faculty.
Within the broader arc of the work, this chapter is crucial because it demonstrates that the syntactic system does not stop at the sentence’s form but extends into its interpretation. The competence of a speaker includes knowledge of how sentences are assigned logical form, and this knowledge is governed by the same kinds of abstract principles that govern sentence structure. By treating LF as a syntactic level, Chomsky unifies the study of meaning with the study of structure, showing that the generative grammar enterprise is not limited to word order and inflection but encompasses the entire system of linguistic knowledge that connects sound to meaning. The Logical Form level becomes the gateway from syntax to thought, and its properties are part of what we must attribute to the mind in order to explain human linguistic abilities.
Chapter 10: The Biological Foundations of Language
We are now in the latter portion of the work, where Chomsky turns from the formal properties of grammar to the question of how language exists in the human species — its biological basis and its place in the natural world.
Chomsky presents the case that language is a biological system, a specific organ of the mind, rather than a cultural invention or a learned skill. He argues that the properties of language we have examined — the deep structures, transformational rules, and principles of Universal Grammar — must be biologically determined, part of the human genetic endowment. The argument proceeds from poverty of the stimulus: the linguistic input available to a child is too impoverished and degenerate to account for the richness and specificity of the knowledge the child acquires. The child does not merely learn language; the child’s mind grows language, much as the body grows arms and legs. Chomsky writes:
“We may think of the language faculty as a fixed function, characteristic of the species, one component of the human mind, which maps experience into a particular state of the language faculty — a particular grammar.”
This biological perspective transforms how we understand linguistic variation. Languages do not differ arbitrarily; they are all realizations of the same underlying Universal Grammar, with variation limited to a small set of parameters that are set by experience. The principles are fixed and universal, part of the species’ biological inheritance. The parameters are the points of variation that the child must determine from the input. For example, a child hearing English sets the head-parameter to head-initial (verbs precede objects), while a child hearing Japanese sets it to head-final (verbs follow objects). The vast range of human languages is generated by a small number of such binary choices, each triggered by minimal exposure to the linguistic environment.
This biological framing matters because it explains why all human languages are learnable by any human child, why languages conform to the same abstract patterns despite enormous cultural diversity, and why language is a species-specific capacity. Chomsky is not merely describing linguistic structures; he is arguing that these structures reflect the initial state of the human mind, the language faculty as a biological organ. The study of grammar thus becomes part of the study of human nature, and the generative grammar enterprise becomes a branch of theoretical biology. The patterns we discover in syntax, phonology, and semantics are not conventions or social agreements; they are the fingerprints of the human genetic code, revealing the structure of the mind itself.
Chapter 11: Language and Mind — The Cognitive Revolution
We have reached the chapter where Chomsky draws together the philosophical and cognitive implications of the generative enterprise, situating his work within the broader cognitive revolution that transformed psychology and philosophy in the mid-twentieth century.
Chomsky contrasts his mentalist approach to language with the empiricist and behaviorist paradigms that dominated psychology before the 1950s. Behaviorism, associated with B.F. Skinner and others, treated language as a set of habits acquired through reinforcement and association. Chomsky’s devastating 1959 review of Skinner’s *Verbal Behavior* is referenced here: he argued that behaviorism could not explain the creative aspect of language use — the fact that speakers produce and understand sentences they have never heard before, in novel situations, without any direct reinforcement history. The behaviorist framework, Chomsky insisted, was not merely inadequate but incoherent as an account of human linguistic capacities. He writes:
“The normal use of language is, in this sense, a creative activity. The speaker’s knowledge of his language enables him to produce and understand an indefinite number of new sentences, many of which bear no point-by-point resemblance to those he has heard before.”
This creative aspect becomes central to Chomsky’s argument that language must be understood as a mental organ with its own internal structure, not as a set of learned responses. The mind is not a blank slate (the tabula rasa of empiricism) but an active, structured system that imposes its own forms on experience. The generative grammar we have been constructing is precisely a theory of this mental structure — an account of the competence that underlies the performance of everyday speech. Chomsky is explicit that his approach is a form of rationalism, a revival of the tradition of Descartes, Leibniz, and the Port-Royal grammarians, who held that the mind contains innate ideas and principles that organize experience.
This matters because it redefines the relationship between linguistics and psychology. Language is not a separate domain from the mind; it is the most accessible window into the structure of human cognition. By studying the principles of Universal Grammar, we are studying the initial state of the language faculty — the genetic blueprint that makes language possible. And by studying how parameters are set and how transformations apply, we are studying the growth of linguistic knowledge in the individual. The cognitive revolution that Chomsky helped lead thus placed language at the center of the human sciences, arguing that the study of the mind must begin with the study of the structures that the mind itself brings to experience.
Chapter 12: The Autonomy of Syntax
Chomsky now turns to a crucial and controversial thesis: the claim that syntax is autonomous — that the principles governing sentence structure cannot be reduced to semantics, pragmatics, or communication. This chapter defends the independence of formal grammar from functional considerations.
The argument proceeds from the existence of well-formed sentences that are semantically anomalous or meaningless. Chomsky’s famous example, “Colorless green ideas sleep furiously,” is a sentence that any speaker of English recognizes as grammatical, despite being semantically nonsensical. The sentence conforms to the syntactic rules of English — it has a subject noun phrase (“colorless green ideas”), a verb phrase (“sleep furiously”), proper inflection, and so on. But it violates no semantic rules; there is no semantic theory that accounts for why these words cannot combine in this way. The grammaticality judgment is purely syntactic. Chomsky writes:
“The sentence ‘Colorless green ideas sleep furiously’ is grammatical, though it is semantically anomalous. Such examples show that syntactic structure is not reducible to semantic structure, and that the rules of syntax must be formulated in terms of formal properties, not meanings.”
This autonomy thesis is further supported by the existence of structural ambiguity — sentences that have two distinct syntactic structures but only one reading, or vice versa. For example, “Flying planes can be dangerous” has two distinct deep structures: one in which “flying” is a gerund and “planes” is its object (that is, the activity of flying planes is dangerous), and another in which “flying” is an adjective modifying “planes” (planes that are flying are dangerous). The ambiguity is syntactic, not semantic — the same surface string maps to two different underlying structures. This demonstrates that syntax has its own generative power, independent of what sentences mean.
The autonomy thesis matters because it restricts the scope of linguistic theory and clarifies what kind of explanation is appropriate for syntactic phenomena. Chomsky is not denying that language is used for communication and meaning; he is arguing that the formal structures that make communication possible have their own internal logic, which must be studied on its own terms. Syntax is not a reflection of semantics; it is an independent computational system that generates structures that are then interpreted by other components of the mind. This separation of syntactic form from semantic interpretation is what allows generative grammar to be a precise, formal science — it studies the combinatorial rules of syntax without needing to invoke the vagaries of meaning or use.
Chapter 13: Language, Thought, and the World
In this chapter, Chomsky explores the relationship between language and thought, and between both of these and the external world. He argues against the view that language simply mirrors reality, and instead develops a more nuanced picture of how linguistic structures relate to human concepts and to the objects we talk about.
Chomsky begins by critiquing the referential theory of meaning, which holds that words stand for objects in the world and that sentences express propositions that correspond to facts. He notes that many linguistic expressions do not refer to anything in the world — for example, “the average man,” “the present king of France,” or “John’s book” (which can refer either to the book John owns, the book John wrote, or the book John is reading, depending on context). These expressions have clear meanings but no simple referents. The deep structure of such expressions often reveals complex underlying relations that are not visible in the surface form. “The average man” is not a person but a statistical abstraction; “John’s book” involves a relation of possession, authorship, or association that must be specified by the grammar. Chomsky writes:
“The relation between language and the world is not a simple one of naming or reference. The expressions of language correspond to highly abstract mental representations, not directly to objects in the external world.”
This leads to a discussion of conceptual structure — the system of mental representations that underlies both language and thought. Chomsky argues that the language faculty maps deep structures to Logical Forms, and that these Logical Forms are the interface between syntax and the conceptual-intentional system of the mind. The concepts we express in language are not given by the world but are constructed by the mind, using innate conceptual categories that are part of the initial state of the language faculty. The world does not come pre-sliced into “objects,” “events,” “properties,” and “relations”; our language imposes these categories on experience. This is why different languages — despite sharing the same Universal Grammar — can carve up the world in different ways at the lexical level.
This matters because it redefines the traditional philosophical problem of how language relates to reality. For Chomsky, language does not directly represent the world; it represents our *thoughts* about the world, and these thoughts are structured by the innate properties of the human mind. The generative grammar enterprise thus becomes part of a larger inquiry into the nature of human cognition — into how the mind constructs its representations of reality. The study of language is the study of a specific cognitive system, not a direct study of the world. And the Logical Form level is the point where syntax gives way to thought, where linguistic structures become mental structures that can be used for reasoning, inference, and communication.
Chapter 14: The Scope and Limits of Linguistic Theory
Chomsky now reflects on what linguistic theory can and cannot explain, setting boundaries for the generative enterprise and distinguishing it from other ways of studying language — such as historical linguistics, sociolinguistics, and language acquisition studies.
He argues that the central task of linguistic theory remains the characterization of competence — the tacit knowledge that speakers have of their language. Performance — the actual use of language in real situations — involves many other factors, such as memory limitations, attention, motivation, and social context, which are not part of the language faculty proper. A generative grammar is a theory of competence, not of performance. It specifies what a speaker knows, not what a speaker does. This distinction is crucial because performance data — slips of the tongue, hesitations, ungrammatical productions — do not directly reveal the underlying competence. To study competence, we need grammaticality judgments, intuitions about sentence structure, and other kinds of evidence that tap into the speaker’s knowledge, not their real-time behavior. Chomsky writes:
“A generative grammar is not a model of a speaker-hearer’s performance. It is a description of the speaker-hearer’s intrinsic competence, the knowledge that underlies actual language use.”
This limitation is not a weakness but a methodological principle. By restricting the domain of linguistic theory to competence, Chomsky makes it possible to develop precise, testable hypotheses about the structure of the language faculty. Sociolinguistics, discourse analysis, and the study of language change are important enterprises, but they operate at a different level of analysis, addressing different questions. They do not threaten the autonomy of generative grammar; they complement it. Similarly, the study of language acquisition becomes the study of how the initial state (Universal Grammar) maps, through experience, to a steady state (a particular grammar). This is a psychological and biological question, not a purely linguistic one, but it is directly informed by the theory of Universal Grammar developed in the earlier chapters.
This matters because it clarifies what is at stake in Chomsky’s program. The generative enterprise is not a theory of everything about language; it is a focused investigation of one specific component of the human mind. By limiting its scope, it gains precision and explanatory power. Chomsky is not claiming that social factors, historical change, or communicative function are irrelevant to language; he is claiming that they are irrelevant to the *formal structure* of the language faculty. That formal structure must be studied on its own terms, using the methods of theoretical linguistics, just as the structure of the visual system must be studied on its own terms, using the methods of perceptual psychology. The autonomy of syntax and the competence-performance distinction are methodological principles that make this kind of investigation possible.
Chapter 15: Language, Freedom, and Human Nature
In the final substantive chapter of the work, Chomsky steps back from technical linguistics to consider the broader implications of the generative theory for our understanding of human nature, human freedom, and the human capacity for creativity.
Chomsky argues that the creative aspect of language use — the ability to produce and understand an infinite number of new sentences appropriate to ever-new situations — is a fundamental property of human beings, one that distinguishes us from machines and from other animals. This creativity is not a matter of random variation or mere novelty; it is rule-governed creativity, the product of a finite system of rules that generates infinite possibilities. The human mind is not a stimulus-response system, nor a simple pattern-matching device; it is a generative system that constructs new structures from a finite store of principles. Chomsky writes:
“The creative use of language, the ability to express new thoughts and to understand entirely new utterances, is a characteristic feature of human intelligence that must be accounted for by any theory of mind.”
This creativity is connected to human freedom. If human behavior were simply a matter of responding to environmental stimuli, as behaviorists claimed, then freedom would be an illusion. But if the human mind has its own internal structures that generate possibilities independently of external input, then humans are not merely passive recipients of experience but active creators of their own mental lives. Chomsky connects this to the rationalist tradition of Descartes and Leibniz, who argued for the innate ideas and mental substance that make human thought possible. The generative grammar enterprise, Chomsky suggests, provides a scientific framework for reviving this tradition, showing that the mind’s internal structure is not a philosophical speculation but a subject for empirical investigation.
This matters because it gives the entire work a humanistic dimension. On Language is not merely a technical treatise on syntax and phonology; it is also a defense of human dignity and human potential. The language faculty is not a trivial curiosity but a central feature of what it means to be human. The study of language reveals the creative, generative, and free nature of human cognition — a nature that cannot be reduced to simple laws of association or social conditioning. Chomsky’s work, for all its formal rigor, is ultimately a contribution to the philosophy of mind and to the understanding of human nature. The generative grammar** is not just a mathematical formalism; it is a model of the human mind’s most distinctive capacity — the capacity to create, to innovate, and to think thoughts that have never been thought before.
