Foundations of Generative Grammar: Structure, Explanation, and the Limits of Behavioral Science
Introduction: The Subject Matter and Aims of Linguistic Theory
We open with Chomsky’s declaration of purpose. This work is not a neutral survey of existing theories but an intervention — a statement of what linguistic theory ought to be. Chomsky locates the central task as the construction of a generative grammar, a system of rules that explicitly and formally assigns structural descriptions to sentences. The aim is not merely to catalog utterances but to characterize the implicit knowledge that enables a speaker to produce and understand an infinite number of novel sentences. This knowledge, which Chomsky calls linguistic competence, is the proper object of inquiry.
From the first pages, Chomsky draws a sharp distinction between two kinds of adequacy a linguistic theory can achieve. Observational adequacy is the minimal standard: the grammar correctly describes the data (e.g., it generates all and only the observed sentences). But a deeper goal is descriptive adequacy: the grammar correctly captures the native speaker’s tacit knowledge — what the speaker actually knows, not just what can be observed. Even more ambitious is explanatory adequacy, which asks why a particular grammar is selected by the child given the data of experience. This tripartite hierarchy — observational, descriptive, explanatory — structures the entire argument.
Chomsky insists that the search for explanatory adequacy forces the linguist to confront the universal grammar that underlies all human languages. Universal grammar is not a list of features common to all languages but rather a system of conditions, principles, and constraints that define the form of any possible human language. This system constitutes the initial state of the language faculty, prior to any linguistic experience. The task of linguistic theory, then, is to characterize this initial state with enough precision to explain how a child, exposed to a finite and often degenerate corpus of speech, acquires a fully articulated grammar.
“The problem for the linguist, as well as for the child learning a language, is to determine from the data of performance the underlying system of rules that has been mastered by the speaker-hearer.”
Chomsky’s move here is transformative. By equating the linguist’s theoretical task with the child’s acquisition problem, he makes linguistic theory continuous with cognitive psychology. The object of study is no longer sentences in the world but the mental representation of language in the mind of the speaker. This reframing — that the data of performance must be explained by reference to an underlying competence — becomes the bedrock on which the entire generative program rests.
The Levels of Adequacy and the Form of Grammars
Chomsky now elaborates the formal requirements for a grammar that meets the standard of descriptive adequacy. A generative grammar must be explicit: every rule must be precisely stated, with no reliance on the reader’s intuition. It must also be recursive, since the set of grammatical sentences is infinite. The core device for achieving recursion is the phrase-structure rule, which expands a category symbol (e.g., S → NP + VP) into a sequence of further categories. These rules generate a phrase-marker, a labeled tree diagram that represents the constituent structure of a sentence.
But phrase-structure grammars alone are insufficient. Chomsky demonstrates that certain grammatical relations — such as active-passive correspondences, question formation, and the interpretation of anaphors — cannot be captured naturally by phrase-structure rules alone without massive redundancy. The problem is that such rules are context-free in their simplest form, and while they generate the right strings, they miss generalizations about how sentences are related. The solution Chomsky proposes is a transformational grammar, which adds a second component: a set of transformations that map an underlying deep structure into a surface structure.
The deep structure is the abstract syntactic representation that determines semantic interpretation; the surface structure is the form that actually appears in speech. Transformations are operations that rearrange, delete, or insert elements in a rule-governed way. For example, the passive transformation moves the object NP to subject position, moves the subject NP into a by-phrase, and inserts the appropriate forms of *be* and *-en*. This single rule captures the systematic relationship between passive and active sentences, whereas a phrase-structure grammar would require two entirely separate sets of rules.
Chomsky then introduces the crucial notion of cyclicity: transformations apply first to the most deeply embedded clauses, then successively to higher clauses. This principle explains why, for instance, question-formation in a sentence like “What did John believe that Mary saw?” applies first to the embedded clause before affecting the main clause. The cyclic principle imposes a strict ordering on rule application, which Chomsky argues is a universal property of human language, not a contingent feature of English.
“A grammar must be, in part, a theory of the deep structures of sentences, and the transformations that relate them to surface structures.”
The importance of this section lies in its technical specificity. Chomsky is not simply suggesting that language has abstract structure; he is providing a precise formalism — phrase-structure rules plus transformations, ordered cyclically — that can be tested against linguistic data. The formalism itself becomes an object of inquiry: which forms of grammar are possible, and why? This sets the stage for the book’s central argument: that the shape of linguistic theory is not arbitrary but is dictated by the biologically given language faculty.
The Aims of Linguistic Theory and the Problem of Induction
Chomsky now deepens the argument for explanatory adequacy by confronting the problem of induction. The child acquires a grammar on the basis of limited, fragmentary, and often error-filled data. Yet every normal child, regardless of intelligence or environment (within broad limits), achieves a rich and highly structured system of knowledge. The only way to account for this, Chomsky argues, is to assume that the child possesses a rich innate schematism that severely constrains the hypotheses that can be entertained.
This argument takes the form of a poverty of the stimulus claim. Consider the formation of yes-no questions in English: a declarative like “The man is here” becomes “Is the man here?” by fronting the auxiliary. But the child encounters data that do not uniquely determine the correct rule. A simpler hypothesis — “move the first auxiliary to the front” — would work for simple sentences but fails for sentences like “The man who is tall is here,” where fronting the first auxiliary would produce the ungrammatical *”Is the man who tall is here?” The actual rule — which requires identifying the main clause auxiliary — is more abstract and structure-dependent. Yet children never seem to entertain the simpler, structure-independent hypothesis. They arrive at the correct rule with no explicit instruction and no negative evidence (they are not told what is ungrammatical).
Chomsky presents this as evidence for the innateness hypothesis: the language faculty contains a specific principle that restricts possible rules to those that are structure-dependent — that is, operations must refer to the hierarchical structure of the sentence, not to linear order alone. This principle is part of the universal grammar that the child brings to the task of acquisition.
“The child must possess a schematism that determines, in advance, the form of the grammars that can be acquired.”
The implications are far-reaching. If the form of possible grammars is innate, then the task of the linguist is not just to describe languages but to discover the universal constraints that make language acquisition possible. Chomsky contrasts this with the empiricist tradition, from Locke to Skinner, which treats the mind as a blank slate on which experience writes. The poverty of the stimulus argument, if valid, refutes that tradition in favor of a rationalist conception of mind, one in which the structure of knowledge is largely determined by the structure of the knowing subject.
Chomsky is careful to state that this does not mean all grammatical properties are innate. The child still must learn which language-specific rules apply (e.g., whether the language is head-initial or head-final). But the *form* that such rules can take, and the principles that govern their interaction, are fixed in advance. This division — between what is universal and what is language-particular — is the central theoretical problem that the remainder of the book will address.
The Structure of the Grammar: Syntactic, Phonological, and Semantic Components
Chomsky now lays out the overall architecture of a generative grammar, dividing it into three major components. The syntactic component is the generative core: it generates deep structures via phrase-structure rules, then applies transformations to produce surface structures. The phonological component maps surface structures onto phonetic representations, interpreting the abstract strings as sound. The semantic component interprets deep structures, assigning meanings. This standard theory (as it came to be called) draws a strong boundary between syntax and semantics: all semantic interpretation occurs at the level of deep structure.
This division is not arbitrary; Chomsky argues for it on both empirical and methodological grounds. Empirically, he cites cases where ambiguous sentences have distinct deep structures (e.g., “Flying planes can be dangerous” has two different deep structures, corresponding to the reading where planes fly and the reading where someone flies planes). Methodologically, the separation of deep and surface structure allows the syntax to be concerned with form, while semantics handles meaning — a division that makes each component more tractable.
Chomsky devotes considerable attention to the phonological component, following the work of Roman Jakobson and Morris Halle. The output of the syntactic component is a surface structure that specifies the morphemes and their order, but not their phonological form. The phonological component applies a set of rules — often ordered and cyclically applied — that convert abstract phonological representations into concrete phonetic realizations. For example, in English, the plural morpheme is realized as /s/, /z/, or /ɪz/ depending on the final segment of the noun. This is not memorized word by word but generated by a rule that refers to the voicing and sibilance of the preceding segment.
“The grammar of a language is a system of rules that specifies the phonetic form and the semantic interpretation of an infinite set of sentences.”
The semantic component is treated more briefly, in part because semantics was less developed at the time. Chomsky adopts a version of Katz-Postal semantics, in which deep structures are directly interpreted by a set of projection rules that assign a semantic representation (a “readout” in a formal semantic marker language). The key claim is that transformations do not change meaning — a principle known as the Katz-Postal hypothesis. This means that all semantic information relevant to interpretation must be present in deep structure. Later developments in generative grammar would challenge this assumption, but within the framework of *Current Issues in Linguistic Theory*, it serves as a working hypothesis that allows the syntax to be studied relatively independently of meaning.
What matters for Chomsky’s larger argument is the modularity of the grammar. Each component — syntactic, phonological, semantic — has its own rules and its own domain of application. They interact in a fixed order: syntax generates deep and surface structures; semantics interprets deep structures; phonology interprets surface structures. This architecture is itself a hypothesis about the mental organization of language, and it is the first detailed proposal for how the human language faculty might be computationally organized.
Deep Structure, Surface Structure, and the Role of Transformations
Chomsky devotes a major section to clarifying the distinction between deep and surface structure, and to demonstrating why transformations are necessary even beyond the passive and question cases already discussed. He introduces the notion of syntactic description as a pair (D, S) where D is the deep structure phrase-marker and S is the surface structure phrase-marker, connected by a sequence of transformations. The deep structure contains all information necessary for semantic interpretation; the surface structure contains all information necessary for phonetic interpretation.
One of the most striking arguments involves discontinuous constituents. In English, a verb-particle construction like “look up” (as in “look up the information”) can be separated: “look the information up.” A phrase-structure grammar would have to generate both orders independently, missing the generalization that they are related. A transformation can derive the separated version from an underlying deep structure where the particle is adjacent to the verb, deleting the particle and moving it rightward. This captures the intuition that “look up” is a single lexical unit, even in its discontinuous form.
Another argument concerns sentence embedding and the phenomenon of extraction. In sentences like “Who did you think that John saw?” the wh-word *who* has moved from its deep-structure position as the object of *saw* to the front of the entire sentence. The transformation that moves it — wh-movement — must apply cyclically, first moving the wh-word to the front of the embedded clause, then to the front of the main clause. Chomsky shows that this cyclic movement explains why extraction from certain positions (e.g., inside a complex noun phrase) yields ungrammatical results: the movement is blocked by constraints that apply at each cycle.
“The transformational rules are, in effect, a set of instructions for mapping a deep structure into a surface structure by a sequence of operations.”
Chomsky introduces the crucial concept of obligatory vs. optional transformations. Some transformations, like the passive, are optional: they can apply or not, producing paraphrases. Others, like the rule that assigns case to noun phrases in English, are obligatory: they must apply for the derivation to converge on a grammatical sentence. This distinction matters because it determines what counts as a well-formed derivation. An optional transformation that does not apply simply leaves the deep structure as the surface structure; a missing obligatory transformation yields an ungrammatical result.
The deepest insight of this section is that transformations are not mere notational convenience. They are a claim about the mental representation of language: speakers represent sentences at multiple levels of abstraction, and the mapping between these levels is rule-governed and psychologically real. The evidence for this comes not from introspection but from patterns of ambiguity, paraphrase, and ungrammaticality that resist explanation at a single level. Chomsky presents this as a cumulative case: the more you look at natural language data, the more you find phenomena that require a two-level (or multi-level) representational system.
Ordering of Rules and the Cyclic Principle
We are now in the section where Chomsky turns from the static description of grammar components to the dynamic question of how rules apply to generate sentences in sequence. This section examines the ordering of transformational rules and introduces the cyclic principle that governs their application.
The organization of rules is not arbitrary. Chomsky argues that transformational rules must apply in a specific linear order, and that this order is itself a part of the grammar that the child must acquire. He presents evidence from English to demonstrate that if rules apply in the wrong order, the grammar will generate ungrammatical sentences or miss necessary derivations. For example, the passive transformation must apply before certain deletion rules; if the order is reversed, the passive auxiliary will be deleted before it can be inserted. The ordering is therefore a constitutive property of the grammar, not a performance convenience.
Chomsky then introduces the cycle as a principle governing rule application in complex sentences. The proposal is that transformational rules apply first to the most deeply embedded S (sentence) node in the phrase marker, then to the next higher S, and so on outward, until the topmost S is reached. At each stage, the rules apply to the domain of that cycle only; once a cycle is completed, its internal structure is no longer accessible to later transformations. This means that transformational operations cannot look back into already-completed cycles. The cycle thus imposes a strict locality condition on syntactic operations.
“The transformational rules apply to a phrase marker in a cyclic manner… first to the most deeply embedded constituent sentence, then to the next most deeply embedded, and so on, until the domain of the highest S in the sentence is reached.”
To clarify: consider a sentence like “John believed that Mary was seen by Bill.” The passive transformation applies first to the embedded clause “Mary was seen by Bill” on its own cycle. Then, after that cycle is complete, the main clause “John believed that S” undergoes its own transformations, possibly including raising or extraposition rules. Because the embedded cycle is closed, the main clause transformations cannot directly manipulate the internal structure of the embedded clause — they can only treat it as a unit. The cycle therefore predicts that certain otherwise possible rule interactions are blocked.
The significance of the cycle is twofold. First, it provides a computationally efficient model of sentence generation: instead of scanning the entire phrase marker at once, the rules work locally on small domains and build up to larger ones. Second, it makes empirical predictions about which sentences are grammatical. For instance, Chomsky shows that the cycle explains why certain reflexivization patterns are possible only within clause boundaries, and why wh-movement can extract a constituent from an embedded clause but cannot extract from within an already-moved constituent. The cycle thus constrains the power of transformations in a principled way.
The Strict Cyclicity Condition and Its Empirical Consequences
This section deepens the analysis of the cycle by introducing the Strict Cyclicity Condition (SCC), a further restriction on how rules interact across cycles. Chomsky is responding to the problem that even with the cycle, some overly powerful grammars remain possible; the SCC eliminates those excesses.
The SCC states that a rule cannot apply in a cycle if it makes use of information that was present only in a higher cycle that has already been completed. Put simply: rules cannot look upward. If a transformation needs to refer to a constituent that is not within the current cycle domain, the rule is blocked. This prevents a situation where, for example, a rule applying in the main clause could somehow alter the internal structure of an embedded clause that has already been processed. The SCC ensures that each cycle is informationally sealed once completed.
Chomsky illustrates this with data from wh-movement and control structures. Consider a sentence like “Who does John want to see?” The embedded clause “PRO to see who” contains a PRO subject (a null pronominal element). Under the cycle, the embedded clause is processed first: the rule of Equi-NP Deletion (or control) deletes the embedded subject under identity with the matrix subject, and then wh-movement applies to the embedded clause, moving “who” to the front. Then on the next cycle, the matrix clause undergoes its own transformations. The SCC prevents the matrix cycle from reaching back into the embedded clause to alter the trace of “who” or to reapply control. This correctly predicts that sentences like “*Who does John want that Bill sees?” are ungrammatical, because the extra complementizer blocks the required movement across a cycle boundary.
“The Strict Cyclicity Condition prohibits a rule from applying in a cycle in such a way that it makes crucial use of information which is not available within the domain of that cycle.”
The empirical payoff of the SCC is substantial. It explains why certain island constraints (such as the Complex NP Constraint and the Coordinate Structure Constraint) hold: they follow from the fact that cycles are sealed and transformations cannot look inside already-completed domains. Chomsky shows that many of the constraints that had been treated as separate, language-specific rules are actually consequences of the SCC plus the cycle. This unification is a major theoretical advance: it reduces the descriptive apparatus while increasing explanatory power.
Chomsky also notes that the SCC interacts with feeding and bleeding relations between rules. A rule order may be feeding if the first rule creates the environment for the second; bleeding if the first destroys that environment. The cycle, combined with the SCC, determines which feeding and bleeding relations are possible across different clause boundaries. This predicts cross-linguistic variation in word order and clause structure, as different grammars may allow or disallow certain rule orders depending on how they instantiate the cycle.
Phonological Representation and the Systematic Phonemic Level
We shift now from syntax to phonology, where Chomsky applies the same logic of formal rule systems to sound patterns. This section introduces the systematic phonemic level of representation, which mediates between the morphophonemic deep representation and the systematic phonetic surface representation. Chomsky argues that a purely phonetic transcription cannot capture the generalizations that speakers know about sound patterns; an abstract phonological representation is necessary.
The systematic phonemic level represents the underlying form of morphemes, including abstract segments that may never surface as such. For example, in English, the plural morpheme is represented as a single underlying form /z/ (the systematic phonemic representation), even though it surfaces as ?” But it cannot do so from certain positions within the embedded clause when that clause has its own subject. The Specified Subject Condition states that no rule can involve X and Y where X and Y are in the same clause or adjacent clauses if there is a specified subject in the clause containing the target of the rule. A specified subject is a lexical subject (not a trace or PRO) that is between the rule’s target and the position being moved to.
“The Specified Subject Condition: A rule cannot involve X and Y if there is a specified subject intervening between them in the domain of the rule.”
Chomsky illustrates with a contrast: “John seems ” or “John seems , it automatically implies certain subcategorization features (such as taking a following NP). The redundancy rules allow the grammar to avoid specifying these predictable features for every lexical entry individually, instead stating the general pattern once. This is an early formulation of what will later be known as lexicalist approaches to morphology and syntax.
What this means for the overall theory is that the grammar’s descriptive power is constrained by the form of lexical entries. The redundancy rules ensure that the lexicon is not merely an arbitrary list but exhibits systematic patterns that reflect the structure of the language. Chomsky emphasizes that these rules are not part of the generative process itself — they do not generate new structures — but rather serve as well-formedness conditions on the lexicon. This distinction will become increasingly important in later work, where the lexicon takes on a more central role in determining syntactic structure.
The Cyclic Principle in Phonology and the Nature of Derived Environmental Conditions
This section returns to phonology to apply the cyclic principle — previously established for syntax — to phonological rules, particularly those that apply in derived environments. Chomsky is here developing the theory of generative phonology first presented in *The Sound Pattern of English* (with Morris Halle, 1968), and he is concerned with the problem of how phonological rules apply to complex derived forms.
The central idea is that certain phonological rules apply only in derived environments — that is, in contexts that arise through the application of morphological or phonological processes, not in the underlying representations of simple morphemes. For example, the rule that palatalizes a consonant before a front vowel (e.g., /k/ → ” or “I want ,” and the repeated NP is deleted under identity with “Bill.”
“The cyclic principle requires that the transformations applying to a complement clause apply before the transformations that affect the matrix clause. This means that the rule of Equi-NP Deletion, which deletes the subject of the complement, must apply on the cycle of the complement clause, and the resulting structure then serves as input to the matrix cycle.”
Chomsky shows that this ordering is not arbitrary but follows from the hierarchical structure of the sentence. The complement clause is an embedded constituent, and the cyclic principle requires that operations on embedded constituents apply before operations on the matrix. This ensures that the complement’s internal structure is fully determined before it interacts with the matrix clause.
The empirical payoff is that the cyclic analysis correctly predicts the distribution of control and raising structures. For example, verbs like “persuade” (which assign a theta role to the object and allow Equi) behave differently from verbs like “seem” (which assign a theta role only to the proposition and allow Raising). The cyclic principle, combined with the Projection Principle and conditions on transformations, provides a unified account of these distinct syntactic patterns. Chomsky is here laying the groundwork for what will later become the Government and Binding theory of control and raising.
The Notion of Grammaticality and the Distinction between Acceptability and Well-Formedness
This section addresses a fundamental methodological issue: the distinction between grammaticalness (well-formedness relative to the grammar) and acceptability (the degree to which native speakers find a sentence natural or understandable). Chomsky argues that these two concepts are often confused in linguistic analysis, leading to faulty generalizations.
Chomsky defines grammaticalness as a property of sentences that are generated by the grammar — that is, they conform to the rules of the syntactic, phonological, and semantic components. Acceptability, by contrast, is a performance notion: it depends on factors such as memory limitations, processing difficulties, and pragmatic plausibility. A sentence may be perfectly grammatical but unacceptable because it is too long, too complex, or violates some processing constraint (e.g., nested center-embedding structures like “The rat the cat the dog chased ate the cheese”).
“The distinction between grammaticalness and acceptability is crucial for the construction of grammars. A grammar may generate sentences that, for reasons of processing complexity, are unacceptable to speakers, but this does not diminish their grammatical status.”
Chomsky illustrates this with the classic example of multiply center-embedded relative clauses. A sentence like “The man who the woman who the dog bit saw left” is perfectly grammatical according to the recursive rules of English syntax, but it is highly unacceptable to most speakers because of the processing load. The grammar does not need to rule out such sentences; rather, the performance system explains why they are difficult.
This distinction matters for the theory because it prevents linguists from confusing performance limitations with grammatical principles. If a sentence is unacceptable but grammatical, the grammar is not at fault — the processing system is. Conversely, if a sentence is acceptable but ungrammatical (as in certain elliptical or colloquial expressions), the grammar must account for the underlying structure. Chomsky’s insistence on this distinction is a recurring theme throughout his work, serving as a defense against empiricist critiques that conflate linguistic competence with performance.
The Form of Phonological Rules and the Notion of Distinctive Features
We now turn to a detailed examination of the formal properties of phonological rules and the system of distinctive features that underlies them. Chomsky builds on the work of Roman Jakobson and others to argue that phonological rules are not arbitrary but operate in terms of a universal set of features defined by phonetic properties.
Chomsky presents the phonological component of the grammar as a system of rewriting rules that map underlying phonological representations (strings of distinctive feature matrices) onto phonetic representations. Each rule takes the form of a structural change applied to a specific structural description, where the structural description is stated in terms of distinctive features. For example, a rule of voicing assimilation might be stated as: “A consonant becomes [+voice] when followed by a [+voice] consonant.”
“Phonological rules are formal operations defined on strings of segments, where each segment is represented as a matrix of distinctive features. The rules apply in a specified order, and their application is governed by the same formal conditions that apply to syntactic rules — the cyclic principle, the strict cyclicity condition, and the derived environment condition.”
Chomsky argues that the distinctive feature system is not merely a notational device but a substantive claim about the nature of phonological representations. Each feature corresponds to a specific articulatory or acoustic property (e.g., [±voice] for vocal cord vibration, [±nasal] for velum lowering, [±continuant] for degree of constriction). The features are universal — all languages draw from the same set — though individual languages may use only a subset.
This is significant because it constrains the form of phonological rules. A rule cannot refer to arbitrary properties of segments; it can only refer to the features that are present in the universal inventory. This reduces the descriptive power of the theory and makes falsifiable predictions about possible phonological alternations across languages. Chomsky’s approach here is a direct application of the same logic that governs the syntactic component: the grammar is not a list of arbitrary rules but a system constrained by universal principles.
The Problem of Abstractness in Phonological Representations
This section grapples with the issue of abstractness — the degree to which underlying phonological representations can differ from their phonetic surface forms. Chomsky argues that the theory of generative phonology allows for highly abstract underlying representations, but this raises methodological questions about how to determine what the correct underlying form is.
Chomsky presents the view that underlying representations are systematic — they are determined by the phonological alternations in the language, not by phonetic surface forms. For example, in English, the plural suffix appears as [s], [z], or [ɪz] depending on the final consonant of the noun (e.g., cats [s], dogs [z], churches [ɪz]). A phonemic analysis might treat these as three separate allomorphs, but a generative analysis posits a single underlying representation /z/ with phonological rules that devoice it after voiceless consonants and insert a vowel after sibilants.
“The abstractness of underlying representations is not a defect of the theory but a consequence of the explanatory goals of generative grammar. The grammar must account not only for the distribution of sounds but also for the regularity of alternations across related forms.”
Chomsky acknowledges the danger of over-abstractness — positing underlying representations that cannot be motivated by any observable pattern. He proposes that underlying representations must be psychologically real: they correspond to the speaker’s internalized knowledge of the language. This does not mean they are directly observable, but they must be consistent with the system of rules that generates the observed forms.
This issue connects to the broader theme of the book: the relationship between data and theory. Just as in syntax, where deep structures are not directly observable, in phonology, underlying representations are abstract constructs that only become visible through the alternations they explain. Chomsky is defending the general methodology of generative grammar: the goal is not to describe surface patterns but to discover the hidden structures that generate them.
The Interaction of Syntax and Phonology: The Role of the Cycle in the Two Components
This final major section of *Current Issues in Linguistic Theory* examines the interaction between the syntactic cycle and the phonological cycle, arguing that the two components are not isolated but share certain formal properties. Chomsky is here synthesizing the work on syntax and phonology presented earlier in the book.
Chomsky observes that both syntax and phonology apply the cyclic principle, but the cycles operate at different levels. In syntax, the cycle applies to syntactic phrases (S, NP, VP, etc.), with transformations applying on each cycle starting from the most deeply embedded constituent. In phonology, the cycle applies to morphological constituents (stems, affixes, words), with phonological rules applying to the output of each morphological cycle. The parallelism is striking: in both components, the cycle ensures that the internal structure of a constituent is fully determined before that constituent interacts with higher-level structure.
“The cyclic principle is a property of the grammar as a whole, not just of individual components. It reflects the hierarchical organization of linguistic structures and the fact that the rules apply to constituents defined by the grammar.”
Chomsky also discusses cases where the two cycles interact. For example, the application of a syntactic transformation can create a derived environment that triggers a phonological rule. Consider the sentence “John will have been reading.” The syntactic cycle generates the auxiliary structure, and the phonological cycle applies to the resulting string. The phonological rules that reduce unstressed vowels or delete certain consonants may apply only to the output of the syntactic cycle, not to underlying forms.
This has a profound theoretical implication: the grammar is not a collection of independent modules but a single system in which the output of one component serves as the input to another. The cycle provides a uniform mechanism for organizing rule application across components. Chomsky ends this section by suggesting that further research will reveal additional properties shared by the syntactic and phonological components, pointing toward a unified theory of linguistic rules.
