Introduction: When a disputed grammatical category is presented as an unquestionable rule
I once watched a TikTok video explaining the difference between 「他学习很好。」 and 「他学习得很好。」 The teacher used 主谓谓语句 (a sentence whose predicate is itself a subject–predicate construction, often called a subject-predicate predicate sentence) to explain the first example. The teacher also remarked that many Chinese-language teachers do not truly understand this construction.
The video then introduced 「他唱歌很好听。」 and 「他唱歌唱得很好听。」, extending a similar explanation to this second pair.
What caught my attention was that the scope and diagnostic criteria of 主谓谓语句 remain disputed in research on modern Chinese grammar.
In their study 《汉语主谓谓语句的鉴别标准问题》 (*The Problem of Identification Criteria for Subject-Predicate Predicate Sentences in Chinese*), published in 汉语学习 on December 15, 2025, Sun Ying (Peking University) and Shi Yuzhi (National University of Singapore) argue that constructions with different grammatical properties have often been grouped under this label. They propose criteria for distinguishing such sentences from topic constructions and adverbial–head constructions.
That study does not automatically invalidate every analysis that uses the category 主谓谓语句. It does, however, raise a methodological requirement: before declaring that other teachers misunderstand a construction, we must establish that our own analysis is supported by the sentence's actual structure.
The questions here are straightforward. How should 「他学习很好。」 be analyzed? Why can't it simply be treated as 「他学习得很好。」 without 「得」? And do 「他学习很好。」 and 「他唱歌很好听。」 really instantiate the same grammatical pattern, or do they merely look alike on the surface?
1. Is 「他学习很好」 correct?
Start with 「他学习得很好。」 — “He studies very well.”
Here, 「学习」 functions as a verb, while 「得很好」 is a state or manner complement describing how well the action of studying is performed.
Two questions make the relations explicit: 「谁学习？」 — 「他。」 (“Who studies?” — “He does.”) and 「学习得怎么样？」 — 「很好。」 (“How well does he study?” — “Very well.”)
Thus, 「很好」 describes the action 「学习」, and 「得」 overtly marks the verb–complement relationship.
But 「他学习很好。」 cannot simply be described as the same construction with 「得」 deleted.
The crucial issue is 「学习」. Alongside its verbal use, the word can have noun-like uses, referring to study or to a person's studies as a domain of evaluation. Guo Xingyan's 2020 research comparing 「学」 and 「学习」 discusses differences in their syntactic functions, including noun-like uses of 「学习」 in certain contexts. That research does not, by itself, establish one exclusive parse for 「他学习很好。」 See the research paper.
One possible analysis of 「他学习很好。」, therefore, treats 「学习」 as a noun-like element denoting the domain of studying, while 「很好」 evaluates his studies or academic performance. Under that analysis, the two sentences may convey similar information without sharing the same syntax: 「他学习得很好。」 evaluates the activity through a marked complement, while 「他学习很好。」 can be read as evaluating his studies.
Similarity in meaning does not establish identity of syntactic structure.
Comparing the two analyses through abstract syntax trees
Borrowing a method from compiler design, a parser (syntactic analyzer) cannot assign two expressions the same abstract syntax tree merely because their meanings are close. It must first determine what function 「学习」 has in each sentence.
For 「他学习得很好。」, 「得」 marks the verb–complement relation, as the tree above shows. For 「他学习很好。」, if we adopt the noun-like analysis, a different tree can be proposed:
This second tree is an analytical hypothesis, not a claim that every occurrence of 「他学习很好。」 must be parsed this way. The contrast illustrates a more limited but important point: we cannot delete a grammatical marker and assume that every syntactic relation remains unchanged simply because the resulting sentence is interpretable.
2. Why can't 「他学习很好」 be used as a template for 「他唱歌很好听」?
Place the two sentences side by side: 「他 + 学习 + 很好」 and 「他 + 唱歌 + 很好听」.
At the level of word order, they appear to share a surface pattern. Surface similarity, however, does not demonstrate a shared grammatical pattern.
In the analysis under discussion, 「学习」 may be noun-like. By contrast, 「唱歌」 has an internal verb–object structure, consisting of the verb 「唱」 (“sing”) and its object 「歌」 (“song”). It also displays the properties of a 离合词 (separable word).
Even if 「学习」 can occupy a particular position in 「他学习很好。」, we cannot simply replace it with 「唱歌」 and conclude that 「他唱歌很好听。」 has the same syntactic relations.
When does abstraction become misleading?
「他 + X + adjective」 is an abstraction: it makes a surface resemblance easier to see. But treating X as a freely interchangeable slot erases the lexical category and internal structure of whatever occupies it. A parser likewise cannot infer that two elements have identical types and internal structures merely because they occur in the same position.
Here, 「学习」 may be noun-like under one analysis, whereas 「唱歌」 is a verb–object construction. Furthermore, 「很好」 and 「很好听」 differ in the kinds of entities they ordinarily evaluate. The same surface pattern does not guarantee the same syntactic tree, much less the same semantic relations.
3. 「她唱歌唱得很好听」: A transparent relationship between the action and its complement
Consider 「她唱歌唱得很好听。」 — “She sings very beautifully,” or more literally, “Her singing sounds very pleasant.”
「她」 is the performer of the action. 「唱歌」 denotes the activity of singing, while 「唱得很好听」 evaluates the manner of singing or the auditory result it produces.
The answer to 「谁唱歌？」 (“Who sings?”) is 「她。」. The answer to 「唱得怎么样？」 (“How does she sing?”) is 「很好听。」.
Because 「唱歌」 is a verb–object construction, the verb 「唱」 can be repeated to take the complement introduced by 「得」. If the object 「歌」 need not be retained, we can also say 「她唱得很好听。」.
In both versions, the intended relation is explicit: 「很好听」 characterizes 「唱」, evaluating the performance or the sound produced by it.
In terms of semantic analysis, 「她」 performs the action 「唱」; 「歌」 is what she sings; and 「很好听」 evaluates the performance or its auditory result. 「得」 makes the connection between that evaluation and the verb visible in the grammatical form.
But in 「她唱歌很好听。」, the form that marks this complement relationship is absent. We therefore need to test the proposed parses instead of assuming that the original complement structure survives unchanged.
4. First analysis: 「她」 is the subject, and 「唱歌很好听」 is a subject–predicate predicate
If 「她唱歌很好听。」 is classified as a 主谓谓语句, the following tree can be proposed:
Under this account, 「唱歌很好听」 must itself form a subject–predicate construction: 「唱歌」 is the embedded subject, and 「很好听」 is its predicate.
The first difficulty concerns the relation between 「她」 and 「唱歌」. If we isolate 「唱歌很好听。」, the phrase does not identify who is singing. In the full sentence, however, the natural interpretation is that 「她」 is the singer.
An action's performer can certainly be recovered from an external constituent or from context; that fact alone does not invalidate this analysis. Still, anyone proposing 「唱歌很好听」 as an embedded subject–predicate construction must explain how 「她」 remains the performer of 「唱」.
The second difficulty concerns the relation between 「唱歌」 and 「很好听」. 「唱歌」 denotes the act of singing, whereas 「好听」 ordinarily evaluates sounds or other things perceived through hearing, as in 「这首歌很好听。」 (“This song sounds good”) or 「她的歌声很好听。」 (“Her singing voice sounds beautiful”).
If 「唱歌」 is directly the subject of 「很好听」, what exactly receives the evaluation: the act of singing or the sound produced by that act?
Assigning the labels “subject” and “predicate” does not, by itself, establish the relation between them.
In parser terms, constructing a tree with those labels does not complete the analysis. We still have to ask: Who sings? What is sung? What exactly is judged to sound good? If the tree identifies 「唱歌」 as the subject of 「很好听」 but leaves the link between 「她」 and 「唱」 unexplained, an important part of its interpretation remains unaccounted for. This is a requirement that the proposed analysis must meet, not an automatic refutation of it.
5. Second analysis: The entire subject–predicate phrase 「她唱歌」 functions as the subject
This possibility must be considered separately because it raises a different semantic issue.
Suppose the entire 「她唱歌」 is a subject–predicate phrase functioning as the subject of the complete sentence, with 「很好听」 as its predicate:
This time the question “Who is singing?” has an explicit answer: 「她」 performs 「唱歌」. But once the entire phrase is promoted to subject position, a different question arises.
5.1. Treating the whole clause as a subject makes the event the object of evaluation
Under the proposed reading, 「她唱歌」 is interpreted as a semantically nominalized event: the fact or event of her singing. This is nominalization at the level of interpretation; it does not necessarily mean that the verb 「唱」 has changed its lexical category to a noun.
On this analysis, 「很好听」 predicates something of the entire event denoted by 「她唱歌」. It is no longer directly a complement of the verb 「唱」.
Replacing the predicate makes the distinction easier to see:
「她唱歌很正常。」 — “It is perfectly normal for her to sing.”
「她唱歌很重要。」 — “Her singing is important.”
In both sentences, 「正常」 (“normal”) and 「重要」 (“important”) can evaluate an action or event directly. But when the predicate becomes 「很好听」, the kind of entity being evaluated changes.
5.2. Does 「好听」 evaluate the event, the performance, or the singing voice?
If 「她唱歌」 is an event-denoting, nominalized subject, the proposed analysis effectively predicates 「很好听」 of the event of her singing.
Yet 「好听」 ordinarily applies to sound. Compare three different relations:
「她唱歌很重要。」 — Evaluates the importance of the event of her singing.
「她唱得很好听。」 — Uses a complement to evaluate how she sings or the auditory result of her performance.
- 「她的歌声很好听。」 — Directly evaluates 「歌声」 (“singing voice”), an auditory entity expressed by a noun phrase.
In 「她唱歌很好听。」, if 「她唱歌」 is the subject of the entire sentence, 「很好听」 must be predicated of an event. To interpret the sentence as “She sings beautifully,” the listener must move from the event of singing to the sound or performance produced by that event.
Natural language allows this kind of semantic shift, but the proposed syntactic tree does not make that shift explicit.
This distinction matters: nominalizing the act of singing into an event does not automatically turn that event into her singing voice. Anyone defending this tree must explain how the interpretation moves between those two kinds of entity.
Thus, 「她唱歌」 as subject and 「很好听」 as predicate is a possible analytical proposal. It is not established merely by showing that 「她唱歌」 is a subject–predicate phrase capable of occupying subject position.
5.3. A syntactic tree can be built while its semantic relation remains unexplained
This is where the analogy with a compiler's semantic analysis becomes useful. The proposed tree places the whole 「她唱歌」 in subject position. Under the nominalized reading, that constituent denotes an event, while 「好听」 ordinarily evaluates sounds or other audible entities. The type of entity being evaluated is therefore not a straightforward match for the predicate's usual semantic selection.
By analogy with type checking, there is a potential mismatch between an event and an auditory entity. To maintain the tree, we must explain how the event of her singing is interpreted as her singing voice or the auditory result of her performance.
Natural language permits such shifts. This is not a rigid programming-language type error, nor does the mismatch alone prove that the original sentence is ungrammatical. It does mean, however, that the proposed analysis has a semantic dependency it must account for. The fact that listeners understand the intended meaning does not make that explanatory step unnecessary.
6. Third analysis: 「唱歌」 is the verbal predicate, and 「很好听」 is its complement
Suppose we retain 「她」 as the subject, take 「唱歌」 as the verbal predicate, and treat 「很好听」 as a complement evaluating the singing. This establishes a straightforward performer–action relationship.
It then faces another issue: the grammatical form of the complement.
The overt form for this relation is 「唱得很好听」. If the verb–object construction 「唱歌」 is retained, the verb can be repeated to form 「唱歌唱得很好听」.
Therefore, if 「很好听」 in 「她唱歌很好听。」 is analyzed as a direct state complement of 「唱歌」, the analysis must explain two things: Why is 「得」 absent? By what rule does the complement attach directly to the verb–object construction?
We cannot assume that deleting 「唱得」 from 「她唱歌唱得很好听。」 leaves all the original syntactic relations intact.
This is why the sentence with an overt complement marker is the clearer teaching example when the goal is to teach the verb + 得 + complement construction. That does not mean every other possible analysis of a sentence without 「得」 must be rejected.
7. Putting the three analyses side by side: Parser → AST → Semantics
A parser receives a sequence of words and proposes possible constituent structures. An AST represents a hypothesis about their relations. Semantic analysis then asks whether that hypothesis accounts for the performer, the singing event, and the entity being evaluated. Applied to 「她唱歌很好听。」, the workflow exposes a distinct issue in each of the three analyses.
If the sentence is divided into 「她」 + 「唱歌很好听」, the tree must explain why 「唱歌」 can function as the subject of 「很好听」 and why 「她」 is nevertheless understood as the performer of 「唱」. Simply calling it a 主谓谓语句 does not supply those missing relations.
If the sentence is divided into 「她唱歌」 + 「很好听」, the performer is explicit. But once the entire 「她唱歌」 denotes an event, the analysis must explain why a predicate ordinarily used to evaluate sounds is predicated of that event. The move from event to singing voice or auditory result requires a semantic account.
If the sentence is divided into 「她」 + 「唱歌」 + 「很好听」, with 「很好听」 treated as a verbal complement, the analysis must justify the complement relationship without 「得」 while accounting for the verb–object and separable properties of 「唱歌」. The tree for 「她唱歌唱得很好听。」, by contrast, displays these relations overtly.
A parser can propose an analysis; it cannot automatically prove that the analysis is correct. The syntactic tree states a structural hypothesis, and semantic analysis tests whether that hypothesis captures the intended relations. The point of the compiler analogy is not to make a computer “rule” on natural-language grammar. It is to prevent recognition of a surface pattern from being mistaken for a completed argument about sentence structure.
8. A construction's occurrence in living language does not automatically make it a standard teaching pattern
「她唱歌很好听。」 can be understood as conveying something close to 「她唱歌唱得很好听。」. But whether a sentence is understood, how its syntax should be analyzed, and whether it can be generalized into a teaching rule are three different questions.
An expression's occurrence in actual language use provides evidence for linguistic research. It does not automatically validate every proposed parse of that expression.
Conversely, difficulties with one analysis do not prove that the sentence cannot occur in the language. Context, ellipsis, and shifts in meaning can all contribute to an interpretation.
The central concern in teaching is overgeneralization from one example. If 「他学习很好。」 is used as a template from which 「他唱歌很好听。」 is assumed to have the same structure, the teacher must demonstrate that the relevant lexical properties of 「学习」 and 「唱歌」 really correspond, and that 「很好」 and 「很好听」 establish the same kinds of relation.
Otherwise, learners may infer an excessively broad and misleading rule: simply place an adjective after a verb to evaluate the action, without examining the verb's internal structure or the adjective's semantic requirements.
Above all, being understandable in conversation does not license assigning an arbitrary parse to a sentence and presenting that parse as a standard construction. A useful teaching rule must explain its syntactic relations and specify the limits of its application.
Conclusion
「他学习很好。」 and 「他学习得很好。」 should not be presumed to be one construction that differs only in the presence of 「得」. The latter visibly marks a complement relationship; the former requires separate examination of 「学习」 and the function it serves in context.
Likewise, 「他学习很好。」 and 「他唱歌很好听。」 may resemble each other in linear word order without sharing the same grammatical structure.
In particular, 「她唱歌很好听。」 presents three different analytical obligations. If 「唱歌很好听」 is an embedded subject–predicate construction, its internal relationship and its connection to 「她」 must be demonstrated. If the entire 「她唱歌」 is the subject, the analysis must explain why 「很好听」 evaluates a semantically nominalized event rather than directly evaluating the singing manner or singing voice. If 「很好听」 is a complement of 「唱」, the account must explain the 「得」 complement construction and the verb–object and separable properties of 「唱歌」.
By contrast, 「她唱歌唱得很好听。」 overtly displays the relationship between the action of singing and its complement. It therefore provides a structurally transparent example for teaching complements marked by 「得」.
From the compiler perspective, abstraction helps identify surface patterns; the parser proposes possible ASTs; semantic analysis checks the relations each tree asserts. These are distinct steps, not interchangeable forms of proof.
A sentence may occur in living language, and listeners may accept an interpretation of it, without either fact automatically establishing a unique syntactic tree or a grammatical rule that can be generalized at will. For a category such as 主谓谓语句, whose diagnostic boundaries remain disputed, the structural label must follow the analysis; it cannot replace the analysis.
References
- Sun Ying and Shi Yuzhi (2025). 《汉语主谓谓语句的鉴别标准问题》. 汉语学习, published December 15, 2025.
- Guo Xingyan (2020). Study comparing the syntactic functions of 「学」 and 「学习」 (full-text PDF).