Linguistic research has long been divided over whether it should adhere to a humanistic standpoint or embrace a scientific, formalized approach. Proponents of the former contend that “linguistics is the study of humans,” stressing the interpretation of linguistic phenomena, the identification of patterns, and critical analysis. Advocates of the latter argue that linguistics should draw on core principles of the natural sciences—such as testability, reproducibility, and cumulativeness—in order to reduce interpretive subjectivity through objectification, datafication, and precision. As computing capabilities continue to expand, this scientific orientation appears to be gaining the upper hand.

American linguist and philosopher Noam Chomsky integrated mathematical logic into linguistic inquiry through his proposal of “transformational-generative grammar,” directing linguistic research toward the study of the mind and seeking to explain the formation of syntax through formal computational methods. His approach quickly rose to prominence and exerted a profound influence on the field for nearly half a century. Yet it also revealed a critical weakness: an excessive emphasis on the scientification of the humanities.

Human understanding and expression do not always operate according to prescribed formulas, nor do people invariably adopt a “1 + 1 = 2” mode of reasoning. The humanities, therefore, should be grounded in principles derived from lived human experience. The “embodied philosophy” and “cognitive linguistics” advanced by American linguist George Lakoff and American philosopher Mark Johnson, among others, inherit Chomsky’s mind-oriented approach to language studies while diverging sharply from his theoretical orientation and methodology. Their approaches emphasize the cognitive mechanisms underlying language formation and explicitly reject formalized methods. In recent years, building on embodied philosophy and cognitive linguistics, China’s localized “embodied-cognitive linguistics” has placed even greater emphasis on a materialist, human-centered standpoint, thereby restoring the authentic nature of linguistic research.

Several years ago, under the strong influence of ideal language philosophy—especially Chomskyan linguistics—the international linguistics community moved toward greater formalization and datafication, at times at the expense of theoretical construction. After this orientation was introduced into China, linguistic statistics, data analysis, and formalized tables came to dominate academic writing, causing many linguistic papers to resemble research outputs in the natural sciences and engineering. Such studies often focus on research questions whose conclusions are already known, and some even lack a clearly articulated argument. Just as this wave of data-centered formalism began to subside, it was quickly followed by a new surge driven by artificial intelligence (AI). As computers, big data, and AI assume increasingly powerful roles in contemporary society, many simple and mechanical tasks can now be delegated to machines and robots.

For example, some believe that translation can now be accomplished with just a few keystrokes, rather than through long hours of human labor. In this context, a new wave of AI is sweeping through the linguistics community: “AI empowerment” has become a buzzword at conferences and lectures, as well as in journals and reports, while MOOCs and microlectures enabled by digital technologies, big-data classrooms, and EEG experiments have gained widespread popularity. Some universities are even considering canceling translation degree programs or suspending foreign-language majors. While such moves may be understandable at non-language-focused institutions, their unchecked expansion risks placing excessive emphasis on the natural sciences at the expense of the humanities. These developments therefore warrant serious attention to avoid falling into the trap of scientism.

Linguistic research should remain grounded in human-centered principles while also drawing on methods involving AI, big data, and experimentation. The humanities and the natural sciences have never existed in a strictly dichotomous relationship; rather, they are closely connected, mutually reinforcing, and complementary. Although advances in modern technology have made human life and work more convenient and comfortable, this does not justify the excessive application of natural science methods in linguistics to the point where the human essence of language is obscured. Accordingly, linguistic research should be grounded in science without being constrained by scientism, and it should pursue theoretical innovation on the basis of empirical data.

Wang Tianyi and Wang Yin are professors from the School of English Studies at Sichuan International Studies University.