There are three different syntactic structures commonly used in Chinese to introduce a new protagonist in narrative discourse: the canonical predicate structure with an indefinite subject (NP-V), the presentational cleft construction (有 yǒu 'there be'-NP-V) and the (stative of eventive) locative-inversion construction (V-NP). According to Zhang (2009), one of the main differences between those three structures is how they are located in the discourse, i.e. in respect to a spatial or/and a temporal point of reference. At the information structure level, the new referent belongs to the focus constituent, and the spatio-temporal constituents are considered as "stage topics" (Erteschik-Shir, 1997). When introducing a new protagonist in narratives, L2 Chinese learners have then (i) to select between spatial or/and a temporal stage topics according to which the focus is introduced; and (ii) to attribute a syntactic position to those two constituents in respect to the verb, knowing that the post-verbal position of the locative-inversion construction seems problematic in L2 acquisition (Yuan 1999). In that context, this study investigates the syntactic and information structures that French-speaking learners of L2 Chinese use to introduce a new protagonist in narrative discourse (Hendriks 1998), by testing two hypothesis: (a) the Topic Hypothesis (Pienemann et al. 2005), according to which, early stage learners topicalize the new NP as an agentive subject (i.e. focus first), while more advanced learners are able to topicalize another constituent (i.e. stage topic first); and (b) the Unaccusative Trap Hypothesis (Oshita 2001), according to which only more advanced learners are able to identify unaccusative verbs in locative-inversion construction and place a new NP in post-verbal position (i.e. focus last). The procedure used consists of the analysis of an oral corpus collected from 5 different groups of informants (N=110): French and Chinese native speakers (NS), as well as learners of L2 Chinese at three proficiency levels (low, intermediate and advanced). The stimulus of the elicited task is a comic strip composed of ten plates of four drawings, including eight target items and two distractors. The narrative is a quest during which the main character meets new protagonists. The results show that 1) the French and Chinese NS use different word order, respectively NP-V and V-NP, but all mostly with explicit spatial stage topic; 2) Low level learners mostly use the canonical word order and rely on implicit chronological stage topic; 3) intermediate learners are not very different from the low level group, even if they show more syntactic diversity; 4) advanced learners show no difference with Chinese NS regarding topic selection and word order. In this respect, the two hypotheses are validated. However, advanced learners show a difference with the Chinese NS regarding the verb, using more stative verbs like 有 yǒu 'there be' and less eventive verbs like 跑出 pǎochu 'run-out', which are more complex and more central in the inaccusative hierarchy (Sorace 1995). Erteschik-Shir, N. (1997) The dynamics of focus structure. Cambridge: Cambridge University Press. Hendriks, H. (1998). Reference to person and space in narrative discourse: a comparison of adult second language and child first language acquisition. SILTA 1, 67-87. Oshita, H. (2001). The unaccusative trap in second language acquisition. SSLA 23(2), 279-304. Pienemann, M. et al. (2005). Extending processability theory. In M. Pienemann (ed.), Cross-linguistic aspects of processability theory, 199-251. Amsterdam: John Benjamins. Sorace, A. (1995). Contraintes sémantiques sur la syntaxe : l'acquisition de l'inaccusativité en italien L2, AILE 5, 79-113. Yuan, B.P. (1999). Acquiring the unaccusative/unergative distinction in a second language: evidence from English-speaking learners of L2 Chinese. Linguistics 37(2), 275-296. Zhang, B.J. (2009). Cóng shī-shòu guānxi dào jùshì yǔyì [From Agent-Patient relation to sentence semantics]. Shanghai: Xuelin Press.