Exploring the effects of animacy and verb type on the processing asymmetry between SRC and ORC among Chinese EFL learners

The study addresses why subject relative clauses (SRCs) are generally easier to process than object relative clauses (ORCs) and examines whether animacy and embedded verb type modulate this asymmetry among Chinese EFL learners. Prior accounts include structure-based theories (e.g., Noun Phrase Accessibility Hierarchy) and memory-based theories (e.g., Dependency Locality Theory), which predict a general SRC advantage. However, evidence shows that properties like noun animacy and verb type can reverse or attenuate the ORC disadvantage, which experience-based approaches such as the Production-Distribution-Comprehension (PDC) account can explain by linking production choices to input distributions and comprehension expectations. Given Chinese RCs are head-final and animacy effects are strong, it is important to test whether Chinese EFL learners show similar patterns and how animacy and verb type interact in both production and comprehension. The research questions are: (1) What are the production patterns of English RCs in Chinese EFL learners, and do animacy and verb type affect them? (2) What are the comprehension patterns of English RCs, and are processing difficulties related to distributional frequency patterns?

Studies consistently report SRCs are easier than ORCs (e.g., Holmes & O’Regan, 1981; King & Just, 1991; Caplan et al., 1998), explained by structural accounts like the Noun Phrase Accessibility Hierarchy (Keenan & Comrie, 1977) and memory-based accounts like the Dependency Locality Theory (Gibson, 1998). Yet findings show animacy modulates RC processing: ORCs with inanimate heads can be easier than those with animate heads (Traxler et al., 2002; Mak et al., 2002, 2006), and such patterns correlate with corpus frequencies (Roland et al., 2007). Experience-based approaches, especially the PDC framework (Gennari & MacDonald, 2009), link production choices to distributional regularities that guide comprehension in a probabilistic manner. Verb type also matters: agent–theme vs. theme–experiencer verbs affect voice choice and argument mapping (Ferreira, 1994; Gennari & MacDonald, 2009), with theme–experiencer verbs more often passivized. Evidence among L2 learners is scarcer, and Chinese presents a typological contrast (head-final RCs), raising questions about transfer and the generality of PDC to L2. Prior mixed results (e.g., Gibson, 2013) motivate testing how animacy and verb type jointly shape distribution and processing in Chinese EFL learners.

Two experiments were conducted. Experiment 1 (Production): - Participants: 35 non-English major undergraduates (21 male, 14 female), age 19–21 (M=20.1, SD=0.7) from PLA Information Engineering University, all right-handed, CET-4 passed (M=589.3/710, SD=61.03). Informed consent obtained; compensated. - Design: Dependent variable = frequency of RC types; within-subject factors = head noun animacy (animate vs inanimate) and verb type (agent–theme vs theme–experiencer). RC types: SRC, ORC, passive RC (PRC; treated as a special SRC to assess voice choice). - Materials: 48 sets of incomplete sentences adapted from Traxler et al. (2002) and Gennari & MacDonald (2008) with familiarized vocabulary. Part A: 36 items (conditions varying head noun animacy and reversing embedded noun–verb order; 12 items with gated completion where only the head noun was given) to test animacy effects. Part B: 24 sets (4 conditions per set) varying verb type (agent–theme vs theme–experiencer) with animate head and embedded nouns; noun–verb order reversed to control order effects. Items intermixed and randomized. - Procedure: Paper-based sentence completion in a quiet classroom; no time limit; independent completion. Non-RCs and ungrammatical RCs excluded. Coding captured RC type, head noun animacy, and embedded verb type; errors irrelevant to RCs ignored. Coding by authors and an experienced English teacher; interrater reliability = 0.96. - Analysis: Frequencies computed; chi-square tests (SPSS 19.0) assessed differences. Experiment 2 (Comprehension): - Participants: 38 college students (22 male, 16 female), age 19–21 (M=20.5, SD=0.5), native Chinese, similar English learning, right-handed, CET-4 passed (M=556.4/710, SD=58.2). Informed consent; compensated. - Materials: Based on Traxler et al. (2005), with familiarized vocabulary. Forty-four sets of SRC/ORC stimuli crossing head noun animacy (animate/inanimate) and verb type (agent–theme/theme–experiencer), yielding six conditions; 60 filler sentences intermixed. Head nouns and embedded verbs matched on frequency and length (no significant differences). Plausibility pretest with N=15 showed no significant difference between SRCs (M=5.66) and ORCs (M=5.83) on a 7-point scale. - Procedure: Word-by-word moving-window self-paced reading (Linger) on laptops. Each trial presented dashes; spacebar revealed successive words with RT per word recorded. Yes/No comprehension question followed each sentence with corrective feedback on errors. Participants instructed to read at normal speed; 10 practice items; session ~1 hour. - Data Processing: Only trials with correct comprehension responses retained. For each participant/condition, RT outliers >4000 ms, <50 ms, or beyond ±2.5 SD per word position removed (avg. data loss 2.13%). For PRCs, function words (by, the) removed to match other conditions. Words segmented into three regions: main clause NP region (the + N + that), embedded clause (NP+V or V+NP), and matrix verb region. Dependent variables: RTs per region and total sentence RT. - Analysis: ANOVAs for region-wise and total RTs; two-way ANOVAs for animacy × sentence type; repeated-measures comparisons for verb type within matched sets; paired t-tests for specific region effects.

Experiment 1 (Production): - Overall frequencies: SRC=572, ORC=420, PRC=325. Adding PRC to SRC yields 897, far exceeding ORC. SRC predominance significant (χ²=70.72, p=0.000), confirming asymmetric distribution. - Animacy effects: With animate heads, SRC=238 (84.7%) vs ORC=43 (15.3%), χ²=135.320, p=0.000. With inanimate heads, SRC decreased and ORC increased; difference significant (χ²=4.693, p=0.030). PRC proportion higher with inanimate heads (23.0%) than with animate heads (10.4%). Frequency order across four RC types: (animate) SRC > (inanimate) ORC > (inanimate) SRC > (animate) ORC. - Verb type effects: Agent–theme verbs: SRC=183, ORC=96, PRC=89; distribution significant (χ²=44.712, p=0.000). Theme–experiencer verbs: SRC=193, ORC=34, PRC=125; distribution significant (χ²=108.483, p=0.000). Theme–experiencer verbs yielded many more PRCs than agent–theme verbs. Experiment 2 (Comprehension): - General SRC vs ORC: SRCs read faster than ORCs. Matrix verb region RTs: SRC=544.12 ms vs ORC=722.80 ms, F(1,37)=52.11, p=0.000. Entire sentence RTs: SRC=4177.41 ms vs ORC=4349.50 ms, F(1,37)=7.22, p=0.000. - Animacy × sentence type: - Entire sentence means: SRC with animate head=4022.71 ms (easiest), ORC with inanimate head=4182.14 ms, SRC with inanimate head=4323.92 ms, ORC with animate head=4526.47 ms (hardest). - Two-way ANOVA: main effect of sentence type significant, F(1,37)=6.04, p=0.014; animacy main effect not significant, F(1,37)=0.08, p=0.774; interaction significant, F(1,37)=8.78, p=0.000. - Simple effects: Animate heads show significant SRC advantage, F(1,37)=17.42, p=0.000; inanimate heads show no significant overall difference, F(1,37)=2.81, p=0.095. For inanimate heads, embedded clause RTs longer for SRC than ORC (2056.72 vs 1923.82 ms), t(37)=2.512, p=0.015. - Verb type: - ORC with animate head: agent–theme total RT=4333.37 ms vs theme–experiencer=4664.21 ms; embedded clause RT difference significant, F(1,37)=6.776, p=0.000 (agent–theme easier). - SRC with inanimate head: theme–experiencer slightly faster than agent–theme (entire sentence 4307.80 vs 4369.71 ms); matrix verb difference not significant, F(1,37)=1.708, p=0.132. Verb type had little effect on SRC due to robust subject preference. - Locus of effects: For L2 learners, differences emerged at embedded clause and matrix verb; stronger reliance on semantic cues (animacy, verb argument structure).

Findings show that Chinese EFL learners produce SRCs far more often than ORCs, and this distribution is modulated by animacy and verb type: animate heads favor SRCs; inanimate heads increase ORCs; theme–experiencer verbs favor passives (PRCs). In comprehension, SRCs are generally processed faster than ORCs; however, animacy can reverse or reduce the ORC penalty (ORCs with inanimate heads become relatively easy), and verb type further modulates difficulty (agent–theme ORCs easier than theme–experiencer ORCs). Crucially, the comprehension difficulties mirror the distributional patterns observed in production, consistent with the PDC account: frequent forms are easier to process because production-driven distributions shape comprehenders’ expectations and mappings between thematic roles and syntactic positions. The effects occur particularly in the embedded clause and at the matrix verb where integration is required, aligning with the idea that verb-mediated argument mapping guides processing. Differences from L1 findings (e.g., locus of reading slowdowns) likely reflect L2 proficiency and chunking differences. Overall, results support experience-based, usage-driven accounts over purely syntactic or memory-only explanations for these animacy and verb type effects among L2 learners.

The study demonstrates that among Chinese EFL learners, SRCs predominate in production and are easier to process than ORCs, but this asymmetry is modulated by animacy and embedded verb type. Inanimate head nouns reduce or reverse the ORC processing penalty, and theme–experiencer verbs increase passive productions and can increase difficulty for active ORCs. Comprehension difficulty patterns closely reflect production-derived distributional patterns, providing strong support for the Production-Distribution-Comprehension account and indicating that prior experience shapes RC processing in both L1 and L2 contexts. Implications for L2 acquisition include leveraging increased exposure to specific RC configurations to facilitate comprehension. Future research should examine additional discourse and referential factors (e.g., sentence complexity, plausibility, indexical reference) and employ advanced methods (e.g., eye-tracking, ERP) to refine understanding of processing mechanisms.

The study focused on animacy and verb type, without systematically manipulating broader discourse factors (e.g., sentence complexity, plausibility beyond pretesting, referential context) or L1 transfer effects. The comprehension method relied on self-paced reading rather than eye-tracking or neurophysiological measures, which may limit temporal resolution and comparability to native speaker studies. The participant sample comprised Chinese EFL learners from a single institution with intermediate proficiency (CET-4), which may limit generalizability. Data sharing is restricted due to participant privacy and institutional confidentiality.