I am a native speaker of Chinese, and I have also used English and German for many years. It is precisely because I know Chinese so well that I have come to feel increasingly strongly that Chinese can, of course, transmit scientific knowledge. It can explain formulas, teach methods, and train people to solve problems. But if someone relies on Chinese alone to form scientific concepts, it becomes extremely difficult to enter the abstract layer of modern science, especially the level at which new scientific concepts and theoretical tools are created.
I need to clarify what I mean by “entering science.” I do not mean reading science textbooks, completing a university degree, or learning to use formulas to solve problems. I mean entering scientific research, particularly the level at which original concepts and theoretical tools are produced.
1. The Three Levels of Scientific Research
I believe scientific research can be divided into at least three levels.
The first level is solving problems within an existing framework.
At this level, a researcher accepts concepts, theories, formulas, and research paradigms that are already available, then uses them to study new objects, solve new data problems, test new phenomena, or improve the precision and efficiency of existing methods.
A person who uses only Chinese can absolutely do this, and can do it very well. Native Chinese speakers do not lack computational ability, logical ability, or the ability to solve complex problems. Once a theoretical framework and its conceptual tools have been built by someone else, a person can master them through training and complete extremely difficult work inside that framework.
The second level is modifying, extending, or generalizing an existing framework.
The researcher is no longer merely using ready-made tools, but begins to change the tools: revising a theory’s domain of application, generalizing an existing structure, introducing new variables or relations, discovering contradictions within a current theory, and expanding the theory into a wider domain.
It is not absolutely impossible for a Chinese-only user to reach this level, but the difficulty has already increased markedly. This level requires more than using concepts. It requires reorganizing the relations among concepts and changing the internal structure of a theory. At this point, the language’s capacity to provide abstract containers begins to matter.
The third level is creating new abstract concepts, formal languages, or theoretical tools.
This level is not about finding a new road on an existing map. It is about creating a new kind of map. It requires a researcher to establish abstract objects that did not previously exist, specify their fundamental relations, and create a theoretical system that can continue to generate questions, derive conclusions, and support later research.
Calculus is a typical example. Before Newton and Leibniz, people had certainly investigated instantaneous change, tangents to curves, areas, and infinite processes. But no one had systematically organized these problems into a conceptual tool that could operate independently and continue to expand. Newtonian mechanics was, to a substantial degree, a systematic synthesis of work by Galileo, Kepler, and others. Calculus came closer to the creation of a new theoretical language and a new cognitive tool.
Yang-Mills theory is important, but under this distinction it belongs closer to the second level: it generalizes the existing Abelian gauge structure to non-Abelian groups and produces new theoretical consequences. This is a profound extension of theory, but it is not the same as creating an entire formal system where no conceptual basis existed before. Saying this does not deny the value of the work. It distinguishes three different levels of originality: solving problems, extending tools, and creating tools.
My judgment is that a Chinese-only user can fully enter the first level; will face clear difficulty entering the second; and will find the third nearly impossible.
“Nearly impossible” is not an absolute proposition that has already been proved by logic. I cannot rule out a future counterexample, nor can I prove that under every possible condition a Chinese-only user could never create a new scientific system. I can only say this: in the history of modern science to date, I have not observed a clear counterexample. I therefore present this as a strong conjecture based on historical phenomena and the structure of the language, not as an unfalsifiable declaration.
2. The First Step of Science Is to Build an Abstract Layer
Science first requires us to extract from concrete objects a relational structure that no longer depends on a concrete image.
Apples, stones, and planets are concrete objects. Mass, velocity, functions, groups, fields, and dimensions are not. They are not things encountered as such in ordinary life. They are abstract objects established through definitions, relations, and rules of operation.
The basic process of scientific reasoning is:
Concrete object → abstract object → formal relation → reasoning and deduction
The decisive point is not the final calculation. It is the first step: can a concrete object be lifted stably into an abstract layer?
Suppose a concept has nominally been abstracted, yet every time the user sees it, the shape of the character, the inherited meaning of the word, and ordinary experience keep pulling the mind back toward a concrete image. In that case, the abstract layer is unstable. I call this phenomenon a “concrete fallback.”
Once a concrete fallback occurs, the user may appear to be employing an abstract concept while actually continuing to understand it through older concepts drawn from everyday experience. The person can memorize a definition, apply a formula, and pass an examination, but has not truly established an abstract object determined only by formal relations.
If a stable abstract layer has not formed at the first step, then however strong the later power of deduction may be, it can only operate within a borrowed formal framework. Strictly speaking, this has not yet reached the level at which scientific concepts are produced.
3. Chinese Lacks Semantically Neutral Containers
The most important difference between Chinese and alphabetic languages such as English is not merely vocabulary, grammar, or information density. It is that Chinese has difficulty providing a clean, semantically neutral, or even semantically blank symbolic container.
Take the simplest expression:
f(x) = yIn this expression, f, x, and y do not need to represent anything concrete in themselves. They may stand for a function, an independent variable, and a result, or they may be redefined according to the needs of a theory. The value of letters lies precisely in their ability to be empty. Their meaning is assigned primarily by definitions and relations, not by everyday meanings that the symbols already carry.
If we replaced them entirely with Chinese characters, we might write something like:
甲 acts on 乙 and equals 丙.
But 甲, 乙, and 丙 are not truly neutral. They already carry associations with order, classification, rank, and traditional usage. They are relatively abstract Chinese characters, yet they are still not semantically blank symbols. When a user sees 甲, the mind does not encounter only a variable position. It may receive its visual form, sound, and inherited cultural meanings at the same time.
When a theory needs dozens of variables, functions, operators, transformations, and hierarchical structures, the problem expands rapidly. Letters can combine freely. They can take subscripts, superscripts, variants, and different typefaces. They can form function names, operator names, and formal systems. They can continue to expand while retaining a low semantic burden.
Chinese characters have much greater difficulty doing this. Modern Chinese cannot freely disassemble radicals and components to invent a completely blank new character for every new concept. Even if a new character were created, it would be difficult to type, print, transmit, and make universally recognizable. Chinese therefore usually has to select existing characters and recombine them into new terms.
The problem is that a Chinese character is not a letter. A letter is a word-building unit with a low semantic burden; a Chinese character is usually a unit with a high semantic burden. When existing characters are combined into a new technical term, old meanings enter the mind before the new abstraction has been established.
Chinese therefore lacks not only a sufficient supply of semantically neutral symbols. Its form is also poorly suited to independently building large systems of operators and abstract formal languages.
4. Mathematical Notation Is Not Pure Chinese
One might reply that Chinese is perfectly capable of teaching mathematics because Chinese textbooks contain functions, equations, calculus, and linear algebra.
But this is precisely where a distinction is necessary. The explanatory language in a Chinese textbook is Chinese; its formal core is often not Chinese.
f(x)=y, dx, ∫, Σ, lim, Latin and Greek letters, and many kinds of operators form a formal language independent of Chinese. Chinese textbooks can teach modern mathematics largely because they embed this non-Chinese symbolic system inside Chinese.
If a person already recognizes and uses Latin letters, and can treat x, y, and f as semantically neutral variable containers, then at the most important conceptual level the person is no longer a strictly Chinese-only user. The person has borrowed a symbolic system from outside Chinese and expanded the available capacity for abstraction.
Therefore, the statement that modern science can be learned “using Chinese alone” often does not describe the actual situation. What is really being used is:
Chinese explanatory language + Western formal symbols + translated scientific terminology
These three components together form the medium of learning. The real question is this: if we remove the Latin letters, Greek symbols, foreign conceptual sources, and modern formal language, leaving only Chinese characters, can Chinese independently construct an abstract system of comparable scale?
My answer is no, or at least that no successful example has yet been observed.
5. The Visual Form of Chinese Characters Causes Concrete Fallback
The interference caused by Chinese does not come only from semantics. It also comes from the visible form of the characters.
For a fluent native speaker, Chinese characters are not transparent signs. A character’s shape, radicals, internal structure, sound, meaning, and accumulated history of use enter cognition together. What the user sees is not an empty container that can be defined arbitrarily, but a visual unit that already has internal structure and a semantic history.
Consider the Chinese term for calculus, 微积分. In mathematics, calculus is a formal theory of limits, rates of change, accumulation, and continuous structures. But when a native Chinese speaker first sees the term, it is easy to read 微 as “minute,” 积 as “accumulate,” and 分 as “divide.”
This reading is not entirely wrong and may even help a beginner remember the term. Yet precisely because it appears explanatory, its everyday meanings can replace the rigorous mathematical definition. The learner acquires the illusion of having understood the concept through the literal meanings of the characters.
The same applies to 秩, the Chinese translation of matrix rank. Matrix rank is a formal object defined through relations such as linear independence and the dimension of the image space. But the character 秩 can lead a Chinese reader toward associations with order, grade, and arrangement. That association has some proximity to the mathematical concept, but it is not the concept itself.
Similar problems arise with the Chinese terms for group, ring, domain, field, manifold, sheaf or layer, and degree of freedom. All of these words have strong and familiar meanings in ordinary language. When they are used in mathematics or physics, it is easy for a user to pull the abstract definition back into an everyday image.
This is concrete fallback: a concept has just been lifted into the formal layer, only to be dragged back into the imagistic layer by the visual form and inherited meanings of Chinese characters.
6. Greater Fluency in Chinese May Not Reduce the Interference
This interference is not necessarily weaker for an excellent native speaker. It may even be stronger.
The higher a person’s command of Chinese, the richer the semantic network likely to be activated by a character. Etymology, allusions, linguistic intuition, rhetoric, collocations, and cultural associations can appear automatically. This is an advantage in literature, history, and everyday expression, but it can become an additional burden when one is trying to construct a rigorous scientific abstraction.
A scientific concept requires the user temporarily to empty a word of its everyday meanings and retain only its formal definition. The more familiar a character is, the harder it is to empty.
The high information density of Chinese characters is therefore double-edged. A small number of characters can carry a large amount of meaning, but precisely because they carry so much meaning, they have difficulty becoming clean abstract containers.
In literature, polysemy can generate rich associations. In science, polysemy can contaminate conceptual boundaries.
7. Translation Cannot Eliminate a Break in Conceptual Genealogy
The main conceptual systems of modern science arose within European languages, mathematical notation, and academic traditions. English, German, and French also use roots with historical meanings, but their scientific terms often took shape gradually within the intellectual development of those languages.
When Chinese receives these concepts, it faces a problem of translation. It must look for approximate counterparts among existing Chinese characters and an existing Chinese conceptual set.
But the original Chinese conceptual set does not naturally contain the objects of modern science. A translation cannot compress the complete history of a concept’s formation, its original problem, its logical relations, and its position within a theory into two or three Chinese characters.
A Chinese user can therefore map an unfamiliar scientific concept onto a familiar Chinese concept. The translated name is remembered, while the relational network in which the original concept was formed may never be rebuilt.
This creates a condition in which lexical familiarity and conceptual unfamiliarity coexist. The more fluent the translated term sounds, the easier it is for the learner to overlook the fact that the term is not an extension of its everyday meaning, but a formal object that must be defined anew.
8. Historical Observation Supports a Strong Conjecture, Not an Absolute Proof
I am not saying that native Chinese speakers lack the ability to create science, nor that Chinese people can solve only simple problems. The question here is whether a person who forms scientific concepts through Chinese alone can enter the third level of scientific research: the creation of new abstract objects, formal languages, and theoretical tools.
To date, I have not seen a clear case in the history of modern science demonstrating that a researcher, relying entirely on a pure Chinese conceptual system, created a foundational modern scientific formal system.
By contrast, native Chinese-speaking scientists who entered high-level theoretical research almost always knew English, German, French, or another foreign language. They also directly used Latin letters, Greek letters, and the formal language of modern mathematics.
Yang-Mills theory did not emerge in a purely Chinese environment. The paper was published in English, the collaborator was Robert Mills, the research environment was in the United States, and the theoretical genealogy, formal language, and system of academic exchange in which the work operated were not purely Chinese.
This does not prove that native Chinese speakers lack originality. It shows that once a native Chinese speaker enters the high-level production of modern scientific theory, the person has in practice also entered a conceptual and symbolic system that extends beyond Chinese.
My position is therefore not:
It has been proved that a Chinese-only user can never create science.
It is instead:
To date, we have not observed a clear counterexample in which a Chinese-only user independently created a third-level system of modern science. We therefore have reason to advance a strong conjecture: the structure of pure Chinese is profoundly unfavorable to the construction of the abstract layer required by modern science.
If a genuine counterexample appears in the future, this conjecture should of course be revised. But until such a counterexample appears, the mere existence of a logical possibility should not be used to avoid the historical phenomena and structural properties of language that are already visible.
9. Pure Chinese Is Better at Transmitting Results Than Producing Concepts
Chinese can transmit scientific results that have already stabilized with great effectiveness.
Once a concept has been defined, a formula established, and a method standardized, Chinese can explain, teach, and popularize it. Through repeated training, learners can master the use of formulas and solve complex problems within an existing framework.
But this does not mean that Chinese is well suited to producing abstract concepts.
Transmitting results and creating concepts are different activities. The first delivers established knowledge to a learner. The second requires the construction of a new abstract object when no ready-made name, structure, or path of thought exists.
A Chinese-only learner can easily learn science as a technique: memorize the definitions, apply the formulas, complete the calculation, and obtain the correct answer under specified conditions. The learner may become highly proficient without truly entering the level at which concepts are generated.
This is not because the learner lacks intelligence. It is because the linguistic tool continually translates abstract objects back into everyday experience and presses formal relations back into the concrete world.
10. Acknowledging the Limits of Chinese Is Necessary to Improve Science Education
To acknowledge that Chinese has limitations in modern scientific abstraction is not to reject Chinese, much less to reject native Chinese speakers.
Language is a tool. To acknowledge that a tool is poorly suited to certain tasks is not to deny all of its value. Chinese has clear strengths in literary expression, historical narrative, the organization of imagery, and high-density communication. Those strengths do not automatically become strengths in the construction of modern scientific formal systems.
The real problem is to mistake “Chinese can translate science” for “Chinese can independently produce the same scientific abstractions.”
If native Chinese speakers are to enter scientific research in the full sense, science education should not attempt to naturalize every concept completely into Chinese. A more reasonable approach is to preserve original terms, mathematical notation, conceptual genealogies, and interfaces with foreign languages, so that a learner understands that a Chinese translation is an entrance, not the concept itself.
Chinese can carry the explanatory layer. It should not be mistaken for a language capable, on its own, of carrying the entire conceptual and formal layers.
Latin letters, Greek letters, and mathematical symbols are not optional habits of notation. They are semantically neutral containers. Once learners master them, they acquire a space in which abstract objects can be established outside everyday semantics.
For this reason, the real task of modern science education is not to prove that Chinese has no shortcomings. It is to supply the formal containers that Chinese lacks.
Conclusion
The problem with pure Chinese is not that it cannot describe scientific results. The problem is that it struggles to provide an abstract space that is sufficiently clean, sufficiently neutral, freely definable, and indefinitely extensible.
The visual form, semantics, and cultural experience carried by Chinese characters make it difficult for them to become completely blank conceptual containers. The moment a new scientific term appears, old meanings have already entered. The moment an abstract object begins to form, concrete fallback may follow.
Yet the first step of science is precisely to build, out of the concrete world, an abstract layer that does not depend on the concrete.
My central judgment is therefore:
A Chinese-only user can solve problems within an existing scientific framework and may be able to modify or extend existing theories to some degree. But relying on Chinese alone to create new abstract concepts, formal languages, and theoretical tools is nearly impossible.
This is not a conclusion that has been proved absolutely. It is a strong conjecture for which no clear counterexample has yet been found.
But the absence of a counterexample is itself a fact worth confronting. Acknowledging this limit does not diminish Chinese. What truly obstructs progress is the refusal to distinguish the cultural value of a language from its capacity to function as a tool of modern scientific abstraction.