A beginner stands in front of a shelf of Chinese textbooks, or scrolls through a long list of apps, with one reasonable question: “Which one is best?” Now place four people beside that learner: one is preparing for a business trip to Shanghai, one wants to read Chinese news, one is studying for the HSK, and one hopes to speak with Chinese relatives. If all four receive the same “best” recommendation, something has gone wrong. Different desired outcomes cannot all be served by the same supposedly optimal resource.
This article argues that the first real decision is not which book or app to choose, but which question to ask first. “What is the best resource?” comes too early. We should begin with: What do we need to do in Chinese?
Everything that follows, including textbooks, apps, media, free learning materials, and tutors, should be selected as a consequence of that answer. The framework below is an editorial and practical model, not a verified research finding. We present it because it can clarify decisions, not because it has been independently established as the universal way to organize Chinese learning.
The Absence of One Correct Path Is Good News
When learners hear that there is no correct textbook or sequence for everyone, they may take it as bad news. It can sound as though the answer exists but has been hidden behind conflicting recommendations, commercial claims, and other people’s success stories. Yet the opposite interpretation is more useful. If learning order depends on intended use, then we do not need to discover a mythical path that is right for all learners. We need to define our own destination well enough to judge the available paths.
Consider four common goals. For travel or basic conversation, listening and speaking may deserve the greatest weight, supported by enough character recognition for the situations the learner expects to encounter. The character load may be much smaller than general advice about “learning Chinese” tends to suggest. That does not make characters unimportant. It means their role is determined by the intended use rather than by an abstract picture of completeness.
For an examination goal, the plan needs to reflect the requirements of the specific test and version being taken. Relevant vocabulary, question formats, assessed skills, and test structure matter because an examination is a defined system. It should not automatically be treated as a neutral measurement of every kind of communicative ability. A resource can be well suited to exam preparation without being the best route to spontaneous conversation, workplace writing, or family interaction.
For work in China, reading, writing, and professional register may move from optional support to the center of the plan. Materials from the learner’s own field can become more useful than broad, general-purpose lessons. For communication within a family or Chinese-speaking community, we must ask about the language variety and interactional setting before choosing materials. Is Mandarin actually the relevant variety? Will communication usually be formal or informal? Will the learner mostly speak and listen, or will reading and writing also matter?
These are not four levels on a single staircase. They are different configurations of listening, speaking, reading, writing, and character knowledge. A learner may also move from one branch to another as work, family, travel, or educational needs change. A goal does not select a textbook automatically. It gives us criteria for asking whether a particular resource develops the abilities we actually intend to use.
Start With a Use Case, Not a Shopping List
A useful goal should be concrete enough for us to imagine a day on which Chinese is being used. “I want to be fluent” is too broad to guide resource selection because it does not identify a situation, task, or output. “I need to follow basic questions and respond during a trip” gives us something observable. So does “I want to read news articles,” “I am preparing for a particular HSK examination,” or “I want to participate in informal family conversations.”
Once the use case is visible, we can turn it into a simple decision tree. What must the learner understand? What must the learner produce? Which forms of Chinese will appear in the real situation? What kind of performance would show that the plan is moving in the intended direction? These questions do not produce a perfect curriculum, but they make resource choices more accountable.
This shift also changes how we interpret difficulty. A resource may be challenging because it develops an ability central to our goal. In that case, the difficulty may be necessary. Another resource may be challenging because it demands skills that are currently peripheral. In that case, effort alone does not make it relevant. The issue is not whether characters, grammar, pronunciation, or writing are inherently worth learning. The issue is when each one must enter the plan, at what depth, and for which intended use.
The same principle applies to sequence. There is no goal-free answer to whether learners should emphasize characters early, delay extensive reading, begin speaking immediately, or spend more time preparing structured output. Each recommendation quietly assumes a destination. Before adopting it, we should identify that destination and ask whether it matches ours.
Five Resource Families, Five Different Jobs
When the goal is clear, the resource question becomes easier because we stop asking which brand is best and start asking what function each resource performs. A practical working classification includes five broad families: textbooks, practice apps, media, free learning materials, and tutors. This five-part classification is not a verified taxonomy. It is a planning tool that helps us avoid expecting one resource to do every job.
A textbook or structured course provides sequence, stable terminology, and a route with identifiable stopping points. Its limitation is that it is normally designed for an assumed learner rather than for one person’s exact circumstances. Even a carefully designed course must decide what to include, what to delay, and how quickly to progress. Those decisions may fit our goal well, partly, or poorly.
Practice apps often specialize in short loops and immediate rewards. That can make them useful for maintaining frequency and returning to selected material. Their weakness appears when the loop begins to stand in for the longer tasks required by the actual goal. Completing prompts is not automatically the same as sustaining a conversation, reading an authentic text, composing workplace writing, or responding under examination conditions.
Media, including podcasts, videos, and articles, can provide real input and expose learners to human voices and connected language. Media cannot, by itself, correct our pronunciation or answer a question tailored to our misunderstanding. Free materials can support reference, explanation, and expansion, but their quality is uneven, so important points may need cross-checking. A tutor offers something especially scarce: individualized feedback and an opportunity for real interaction. The cost is equally clear, whether measured in money, scheduling, or the work required to find a suitable match.
Seen through function, an apparent contradiction disappears. One resource can be excellent for explaining grammar and nearly useless for creating conversation practice. Another can produce abundant listening input without offering any structured sequence. If a learner studies with an app for months and still cannot speak, the problem may not be that the app failed at its actual function. The problem may be that a tool built for short practice loops was expected to create opportunities for dialogue.
Build One Backbone, Not a Collection
Our preferred working principle is simple: build around one backbone rather than accumulating a collection. In most resource configurations, one textbook or structured series should hold the main sequence and keep terminology reasonably stable. One reference tool should serve as the default place to check words, characters, or questions. Additional resources should then have explicit supporting roles.
Listening material should be added because the backbone does not provide enough sound, not merely because a podcast is popular. Speaking tasks should be added because knowledge needs to be tested through output, not because every learner is supposed to use the same tutor or platform. Professional texts should enter when workplace reading or writing is part of the intended use. Exam materials should enter when a specified examination is the target. Each addition should answer a visible gap.
This model responds to a familiar practical failure: app switching and bookmark accumulation. A beginner downloads several applications, saves many free resources, and moves between them without completing a sequence. We should not automatically interpret this as a lack of discipline. If no backbone has been chosen, every new resource appears equally valuable, and none receives enough sustained attention to show what it can contribute.
A backbone does not require rigid, uninterrupted linear study. We can return to earlier lessons for reinforcement, add speaking tasks when knowledge remains trapped on the page, or reduce the immediate character load when conversation is the near-term priority. The backbone coordinates the plan. It does not forbid adaptation.
This distinction matters because structure and flexibility are not opposites. Without structure, changes can become random switching. Without flexibility, structure can continue long after it stops serving the goal. A workable plan gives the learner a default path and clear reasons for departing from it.
Change the Plan When the Output Disappears
New resources appear constantly, but novelty alone is not a sufficient reason to restart. A more defensible review begins with observable changes. The first is a change in the intended use of Chinese. A promotion, a change of profession, or a move to a different place can alter the balance among speaking, reading, writing, specialized vocabulary, and register. When the destination changes, the resource configuration may need to change with it.
The second signal is that the plan does not contain the main task required by the goal. A conversation-oriented learner who is never required to listen and respond has a structural gap. A reading-oriented learner who never encounters connected, authentic texts has another. We can make substantial progress inside a resource while still avoiding the performance we originally wanted to develop.
The third signal is an inability to produce a testable output corresponding to the goal. The relevant product will differ by branch. It may involve responding in a basic interaction, working through the tasks of a particular exam, reading material connected to professional practice, or participating in a family conversation. The purpose is not to impose a single universal assessment. It is to check whether study activity is connected to intended use.
These signals are more informative than a vague sense that another book might be better. We should change a plan when the present structure no longer serves the desired output, not simply because a new recommendation has appeared. The key question is not “Is this resource good?” but “What role does it play, and is that role still necessary?”
Conclusion and Limits
There is no textbook or learning order that is right for everyone because “right” depends on what the learner intends to produce with Chinese. The first decision is therefore not a purchase. It is a sufficiently specific description of use: what we expect to understand, say, read, or write in a recognizable situation.
From that description, resources can receive distinct jobs. Textbooks provide structure, apps support short practice loops, media supplies connected input, free materials extend reference and explanation, and tutors provide feedback and interaction. One structured resource can act as the backbone, while other materials fill gaps that the backbone does not cover. When the goal changes, the central task is missing, or no relevant output can be produced, the plan deserves review.
This article does not settle which textbook is best, how many hours learners should study, which brands are most effective, or what completion rates different programs achieve. It does not establish the proposed decision tree or five-family classification as research findings. Those claims are not independently verifiable from the evidence available for this article.
The framework should therefore be treated as an analytical tool, not a guarantee of learning outcomes or a universal prescription. Its value must be tested against each learner’s circumstances. Being explicit about that limit is more useful than filling the evidence gap with a confident ranking that the available sources cannot support.
Sources cited
- Matrix Hanzi editorial essay, “Mục tiêu trước, tài nguyên sau”