Against Deep Structure: A Saussurean Reconsideration of Generative Linguistics

Introduction

During the 1960s and 1970s, generative linguistics introduced the influential distinction between surface structure and deep structure, a framework intended to explain how sentences are generated from underlying syntactic representations. Sentences that appear different on the surface—such as active and passive constructions—were understood as sharing a common deep structure representing their core semantic relations.

Ferdinand de Saussure approached language from a markedly different perspective. In the Course in General Linguistics, he emphasizes the systemic organization of language, arguing that linguistic elements acquire value only through their relations within the whole system. Roy Harris (2003) later questioned the generativist critique of Saussure, suggesting that abandoning observable linguistic criteria in favor of hypothetical underlying structures creates serious methodological difficulties. From this perspective, Saussure’s systemic conception of language may provide a more stable framework for linguistic analysis than the deep-structure hypothesis.

Generative Grammar: Surface and Deep Structure

Generative grammar proposes that sentences exist at two distinct levels of organization. Surface structure corresponds to the form of an utterance as it appears in speech or writing, while deep structure represents an underlying configuration encoding the fundamental semantic relations of the sentence. Transformational rules link these levels, allowing different surface forms to derive from the same abstract structure.

A well-known illustration involves the relationship between active and passive constructions. The sentences The dog bit the postman and The postman was bitten by the dog differ grammatically but are assumed to originate from the same deep structure. Transformational operations rearrange syntactic constituents while preserving the underlying semantic roles.

Roy Harris (2003) describes this framework as relying on a “dogmatic distinction” between surface and deep levels. From the generativist perspective, Saussure’s failure to posit such an underlying structure appeared to be a limitation of structural linguistics.

Saussure and the Systemic Nature of Language

Saussure’s linguistic theory begins not with isolated words or sentences but with the system of language as a whole. Language, in his view, is not a collection of independent elements but a structured network of relations. As he writes, “Language is a system of interdependent terms in which the value of each term results solely from the simultaneous presence of the others” (Saussure, 1916/2011).

Within this framework, linguistic units derive their value from contrasts and oppositions within the system. Saussure summarizes this relational principle with the claim that “in language there are only differences without positive terms” (Saussure, 1916/2011).

This systemic perspective also determines how linguistic units are identified. Segmentation cannot proceed from isolated words or expressions but must be grounded in the relational structure of the language itself. Units emerge through their syntagmatic and associative relations within the system. Rather than searching for hidden layers beneath observable forms, Saussure directs linguistic analysis toward the network of relations that makes those forms intelligible.

Harris’s Critique and Saussure’s Vindication

Harris argues that once linguistics abandons the observable criteria used to delimit linguistic units, the range of possible underlying structures becomes virtually limitless. If sentences with different grammatical forms can share the same deep structure, the same reasoning might be extended to verbs and adjectives, or to nominalizations derived from verbal expressions. Without clear criteria for delimiting units, the number of possible deep structures becomes indeterminate.

In later developments within generative linguistics, the centrality of deep structure has been reduced. Steven Pinker (1994), for example, notes that the concept has increasingly been treated as a technical device rather than as a direct representation of cognitive reality.

From this perspective, Saussure’s approach appears strikingly prescient. By grounding linguistic analysis in the relational structure of the language system, his framework avoids the proliferation of hypothetical underlying levels. The systemic organization of language provides clear criteria for identifying linguistic units and understanding their interactions.

Conclusion

The generative distinction between surface and deep structure initially promised to reveal the mechanisms underlying sentence formation. Yet the search for deeper levels of representation introduces problems of theoretical indeterminacy. Saussure’s structural conception of language offers a different approach. By treating language as a system of relational differences, it locates the explanation of linguistic phenomena within the organization of the system itself rather than in hidden structures beneath observable forms.

Although generative linguistics sought to move beyond Saussure, later critiques suggest that his systemic insight remains a powerful foundation for understanding how linguistic structures function.

References

Harris, R. (2003). Linguistics after Saussure. In P. Cobley (Ed.), The Routledge Companion to Semiotics and Linguistics (pp. 243–259). Routledge.

Pinker, S. (1994). The Language Instinct: How the Mind Creates Language. William Morrow.

Saussure, F. de. (2011). Course in General Linguistics (W. Baskin, Trans.). Columbia University Press. (Original work published 1916)

---