The role of L1 and L2 in the acquisition of null subjects by Chinese learners of L3 Italian

ABSTRACT

This study investigates the acquisition of null subjects by Mandarin Chinese native speakers (‘Chinese NSs’ hereafter), with different levels of L2 English proficiency, at initial stages of L3 Italian acquisition. The aim is to find out if, when acquiring Italian null subjects, Chinese NSs resort to their L1, which, like the L3, allows null subjects, or to the L2, that does not allow null subjects but may be perceived as typologically closer to the L3, given the shared affiliation to the Indo-European family. For this purpose, four groups of Chinese NSs all at the same stage of L3 acquisition but differing with respect to their L2 English proficiency level, were tested. The results of an acceptability judgement task showed that the lowest L2 proficiency group, in two conditions (i.e. declaratives and wh-questions), was the most accurate in rejecting overt subjects, although in the third condition (i.e. expletives) they did not perform as accurately. However, the overall high acceptance of null subjects suggests that L2 English did not act as the sole source of transfer for learners with a higher proficiency: L1 transfer seems to contribute to shaping L3 acquisition too, possibly combined with learning-environment-related factors.

KEYWORDS:

• Null subjects
• L3 acquisition
• Italian
• Chinese
• pro-drop
• topic-drop

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Null subjects (PRO) are allowed in non-finite embedded clauses, bound by a matrix argument. Also, in some genres, e.g. diary-style, clause-initial subject omission is acceptable if the referent is obvious.

2 Newell (1990, p. 136) describes automatic processing as fast, unstoppable, exhaustive, unconscious, and independent of the information load.

3 I will not discuss predictions based on CEM, L2SFM, and L1 transfer hypothesis, since, as mentioned in the previous section, existing empirical evidence seems to put them out of contention as tenable models of L3 acquisition at the initial stages (see Puig-Mayenco et al., 2020, for a review).

4 The Common European Framework of Reference is an international language ability standard using a six-point scale, from A1 (beginner) to C2 (near-native).

5 For Matrix and Embedded conditions, the independent variable of Matrix (i.e. whether the null/overt subject appeared in the matrix/embedded clause) was initially included. The model was progressively backwards-fit and non-significant fixed effects and interactions were removed; the term Matrix was consequently removed.

6 Following Barr et al. (2013), random-effect structures were kept as maximal as possible in all models, in all conditions.

7 Unnatural and even ungrammatical “teacher talk” is reported by several scholars (e.g. Hakansson, 1986; Ishiguro, 1986; Kleifgen, 1985). Kleifgen (1985) explains teachers’ use of ungrammatical or unnatural utterances as a form of adaptation to the learners’ proficiency and needs.