Description

Sentence comprehension is a complex process. Researchers have attempted to identify universal parameters that constrain language comprehension cross-linguistically by investigating subject (SRC) and object relative clause (ORC) comprehension. One hypothesized parameter is a universal processing advantage for SRCs over ORCs, which has been supported by research involving languages that exhibit head-initial relative clause structures (e.g., English). However, findings are much less consistent in confirming a universal SRC advantage within languages with head-final relative clause structures (e.g., Mandarin). The purpose of the current study was to test the universality of the SRC advantage and investigate canonical and noncanonical sentence processing in Mandarin (canonical: actives and ORCs; noncanonical: passives and SRCs) and English (canonical: actives and SRCs; noncanonical: passives and ORCs). Nineteen older Mandarin-English bilinguals (Meanage=77, SDage=4, Rangeage: 70-86; 11 female) were administered the English SOAP (A Test of Syntactic Complexity) sentence-picture matching task and an adapted Mandarin version (M-SOAP) to test comprehension of actives, passives, SRCs, and ORCs across both languages. Paired t-tests revealed no difference in performance of SRCs and ORCs in Mandarin. However, participants performed significantly better on English SRCs over ORCs [t(17) = 5.584, p < 0.0001, Cohen’s d = 1.694]. Results from a 2x2 repeated measures ANOVA yielded significant main effects of language [F(1,17) = 40.084, p < 0.0001] and canonicity [F(1,17) = 32.628, p < 0.0001], and a significant language×canonicity interaction [F(1,17) = 20.842, p < 0.001]. Participants performed worse on noncanonical (accuracy=71%) than canonical (accuracy=87%) sentences, and this effect was driven by a significant noncanonical-canonical difference in performance in the non-dominant English language [t(17) = 5.609, p < 0.001, Cohen’s d = 1.256]. With many participants achieving high scores in their dominant language, the performance difference on Mandarin canonical and noncanonical sentences remained insignificant. However, closer examination revealed performance variability where lower M-SOAP scores were associated with lower noncanonical relative to canonical scores (r=-0.68, p=0.001). These findings provide evidence against a universal SRC processing advantage and demonstrate that sentence processing is driven by the canonicity of sentences.