Syntactic representations encode grammatical functions: evidence from the priming of mapping between grammatical functions and thematic roles in Cantonese

ABSTRACT

A structural priming study examined whether the mapping between grammatical functions (e.g. subject) and thematic roles (e.g. agent) persists across utterances. Specifically, we tested whether native Cantonese speakers are more likely to produce a standard subject–verb–object (SVO) active or a standard passive sentence after repeating an active sentence in object–subject–verb (OSV) order where the direct object (i.e. patient) is grammatically marked as a topic. Such Topic-OSV and standard active sentences share the same agent–subject and patient–direct object mapping, but they differ in all other important aspects including constituent structure, information structure, and linear thematic role order. In contrast, Topic-OSV and passive sentences only share information structure and thematic role order. The results demonstrated that Topic-OSV prime sentences primed SVO actives rather than passives, suggesting that grammatical function–thematic role mapping is a major source of structural priming. We thus argue that our sentence representations encode grammatical functions.

KEYWORDS:

• Structural priming
• grammatical function
• language production
• sentence processing
• Cantonese

Data availability statement

The data of this study are available at https://osf.io/dgye7/.

Disclosure statement

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

Notes

1 perf refers to perfective aspect, which indicates completion of actions.

2 cl and sfp refer to noun classifier and sentence final particle, respectively; numbers indicate lexical tones.

3 top refers to sentence topic.

4 Note that intuitively, priming from Topic-OSV/passive to Topic-OSV/passive constructions is most conceivable. However, the results of our pilot study showed that at least in our experimental setting, Cantonese speakers do not use the topic particle ne1 in isolated, out-of-context sentences. They did not use OSV constructions without the topic marker either in the pilot study. Our main study replicated all these results. See the Discussion section for a discussion regarding this issue.

5 As an anonymous reviewer suggested, the priming from the Topic-OSV to the standard SOV active construction may be at least partly attributed to the similarity between the constituent structures of these two constructions: Note that the constituent structures both contain the NP–V (or [NP [_VP V]]) structure, although they noticeably differ as a whole ([ NP [_S NP [_VP V]]] vs [_S NP [_VP V NP]]). This possibility must await further research, as previous structural priming studies have traditionally considered whole constituent structures, rather than only part of them, the main entity of constituent structure priming.