The present study investigates transfer in L3 acquisition, testing the Cumulative Enhancement (Flynn et al. 2004) and the Typological Primacy (Rothman, 2011) models through an investigation of French adverb placement and object clitics. I compare proficiency-matched L3 and L2 French speakers whose prior languages are L1-Sinhala–L2-English or L1-English. In French, adverbs may appear between the finite verb and the direct object. However, French does not allow adverbs between the subject and the finite verb, whereas, in English, they may. In English, adverbs may not intervene between the finite verb and the direct object, unlike in French. Sinhala, on the other hand, allows both preverbal and postverbal adverb placement. In French, object pronominalization is realized through preverbal clitics. Sinhala also has preverbal object pronouns but additionally allows null pronouns, which are ungrammatical in French. English has only overt postverbal object pronouns. The Cumulative Enhancement Model proposes that the grammar of previously acquired languages enhances subsequent language acquisition. Under this model, the L3 speakers would be more target-like on postverbal adverbs and preverbal clitics than the L2 speakers, due to facilitation from Sinhala. The Typological Primacy Model proposes that the structurally more similar language transfers in L3 acquisition. Under this model, both groups would experience transfer from English, so their performance would be similar. The data was collected via an audio acceptability judgement task, a written acceptability judgment task and a production task. The results for adverb placement support the Cumulative Enhancement Model. However, the results for object clitics suggest that the L2 group is more target-like than the L3 group. This suggests that the results are not compatible with either of the two models. Therefore, I discuss the results in relation to L3 accounts that predict negative transfer from any previously acquired language (Slabakova, 2016; Westergaard et al., 2016).
