Abstract
This paper investigates the syntactic derivation of the non-canonical object construction (NOC) in Mandarin Chinese. By analyzing the transitivity of verbs, we examine two primary cases: NOCs with transitive verbs and those with unergative verbs. Utilizing the theories of Phase and Phase Extension, we argue that the unique hybrid properties of non-canonical objects—exhibiting traits of both direct objects and prepositional phrase (PP) objects—stem from a dual licensing mechanism involving both a null preposition and the verb.
The Puzzle of Non-Canonical Objects
In standard linguistic theory, verbs typically select a "Theme" as their direct object (e.g., eat an apple). However, Mandarin Chinese frequently employs the Non-Canonical Object Construction (NOC), where the post-verbal position is occupied by locations, instruments, or goals. Examples include chi shitang ("eat canteen") instead of the standard zai shitang chi ("eat at the canteen"), or xie maobi ("write writing brush").
The central challenge lies in explaining the "licensing" mechanism: how does a location or instrument, usually requiring a preposition, come to behave like a direct object? Previous accounts often relied on "deleted prepositions" without explaining the structural steps or why these objects retain certain PP-like behaviors while gaining object-like syntactic positions.
Theoretical Framework: Phase Extension
This study adopts the framework of Generative Grammar, specifically focusing on Chomsky’s Phase Theory and Bošković’s (2014) Phase Extension Theory. In standard Phase Theory, syntactic structures are built in discrete chunks (phases); once a phase is completed, its internal contents are "frozen" and inaccessible for further movement.
Bošković’s Phase Extension offers a dynamic alternative: if the head of a phase (like a preposition) moves into a higher head (like a verb), the boundaries of that phase "extend" upward. This extension creates an "escape hatch," allowing elements previously trapped inside a PP to move into higher structural positions reserved for direct objects.
Deriving NOCs: Transitive and Unergative Verbs
The paper demonstrates that NOCs are derived through a process of Head Movement. The non-canonical object originates inside a PP headed by a null (invisible) preposition. When this null preposition moves to merge with the verb (P-to-p-to-v), the phase is extended.
Case 1: Transitive Verbs
For transitive verbs like chi ("eat"), the object shitang ("canteen") originates as the complement of a null preposition. As the preposition moves to the verb, the phase extends, allowing shitang to move to the Spec-vP or a related higher position. This explains why the location precedes certain low-level modifiers but follows the main verb.
Case 2: Unergative Verbs
Unergative verbs, such as fei ("fly"), involve an agent but no standard object. In the NOC fei Shanghai ("fly to Shanghai"), the goal Shanghai is licensed through the same phase extension mechanism. The movement of the null preposition into the verb allows the goal to escape the PP and occupy a position where it can be syntactically licensed as an object.
Key Insights and Results
- Dual Licensing: The "hybrid" nature of these objects is explained by their origin: the null preposition provides the semantic role (Location/Goal), while the verb provides the syntactic licensing (Case) required for the object position.
- Structural Validation: The model accounts for why unaccusative verbs (e.g., lai "come") cannot form NOCs, as they lack the necessary "Voice" or light verb structure to trigger phase extension.
- Theoretical Reconciliation: This approach reconciles the debate over whether these are "true" objects or adverbials by showing they are structurally both: they begin as adverbial PPs but transform into syntactic objects through movement.
Conclusion
By applying Phase Extension Theory, this paper provides a unified account for the derivation of NOCs in Mandarin. While the model successfully addresses the syntactic requirements, future work remains to integrate prosodic constraints (such as the monosyllabic preference in NOCs) and to explore whether this mechanism explains similar oblique-to-object shifts in other languages.
Summary updated