Language Has Never Had a Single Theory, and That Is the Point
The history of linguistics is often told as a tidy sequence: structuralism gave way to generative grammar, which gave way to cognitive and social theories. It is an elegant story. It is also misleading.
Linguistics has never progressed in a straight line. It has moved in cycles, each generation isolating one truth about language, elevating it into a complete theory, and then discovering what that theory had excluded. The field does not evolve by accumulation. It evolves by oscillation.
The real tension has always been the same: is language a system or an activity, a mental structure or a social practice, a formal code or lived experience?
Every major tradition answers by narrowing the frame. And every subsequent tradition reopens it.
The modern story begins with the idea that language is a structure of differences rather than a collection of words. In this view, meaning does not reside in things themselves but in their relations within a system. This move made linguistics scientific. It also made it strangely weightless. The speaker disappeared, replaced by an abstract system that seemed to operate without human presence.
The correction came almost immediately. Other linguists insisted that structure cannot be separated from purpose. Sounds exist not as abstract units but as tools for distinguishing meaning. Sentences are not neutral forms but carefully organized flows of information. Language, in this view, is not only structure; it is communication design.
Yet this functional turn created its own blind spot. Once language is defined entirely by use, it becomes difficult to explain why it has stable internal regularities at all.
At the opposite extreme, some scholars attempted to rescue precision by stripping language of everything except its formal architecture. Meaning, psychology, and even physical sound were treated as secondary. What remained was a purified system of relations, elegant, internally consistent, and increasingly detached from how people actually speak.
That detachment triggered another reversal. In mid-century American linguistics, researchers insisted that only observable behavior should count as scientific data. Language was treated as an empirical object to be segmented, classified, and described. The result was extraordinary descriptive discipline but a shrinking ability to explain creativity, ambiguity, and the human capacity to generate endless new utterances.
The break came when language was relocated inside the mind. Under this view, language is not primarily social behavior or external structure but an internal cognitive system. Humans are not simply trained into language; they are biologically equipped for it. This shift transformed linguistics into a branch of cognitive science, but it also introduced a different abstraction: the ideal speaker, stripped of context, variation, and real-world messiness.
Unsurprisingly, the reaction was to bring the body and the world back in. Language, critics argued, cannot be reduced to an internal computational system. It is grounded in perception, metaphor, and embodied experience. We do not merely process grammar; we think through it. Meaning is not computed in isolation; it is shaped by how humans physically and conceptually inhabit the world.
Others pushed the argument further in a different direction: the speaking subject itself cannot be ignored. Words like “I,” “you,” “here,” and “now” do not have fixed meanings. They acquire meaning only in the act of being spoken. Language does not merely express identity; it produces it in real time. The speaker is not outside the system. The speaker is generated by it.
Still another expansion refused to treat language as an isolated object at all. From this perspective, language is only one layer of a much larger semiotic environment. Myths, literature, rituals, and cultural systems form an interconnected web of meaning-making. Language is not the boundary of communication; it is one node in a broader cultural intelligence system.
Seen together, these traditions do not form a ladder of progress. They form a pattern of correction. Each time linguistics becomes too abstract, it is pulled back toward usage and context. Each time it becomes too empirical, it is pushed toward deeper structure. Each time it reduces language to mind, it is reminded of culture. Each time it reduces language to culture, it is reminded of cognition.
This is why no single theory of language has ever survived for long without modification. Each contains a genuine insight, and each becomes unstable the moment it claims completeness.
The same unresolved tension is now visible in artificial intelligence. Large language systems learn structure from massive datasets, infer patterns from usage, and generate output that resembles linguistic competence without fully explaining what language is. They do not resolve the contradictions of linguistic theory; they inherit them.
What emerges from this long intellectual history is not a unified science of language but something more uncomfortable and more accurate: language resists final explanation because it is not one thing.
It is at once a system of structure, a tool of communication, a cognitive capacity, a social practice, and a cultural medium. Any theory that fully captures one dimension tends to distort the others.
The deeper lesson, then, is not that linguistics has failed to find the right model. It is that language itself does not permit one. It continually exceeds the boundaries of the frameworks built to contain it, and that is why linguistics does not converge. It oscillates.
