Conventionality and context jointly modulate the effect of inhibitory control on L2 metaphor comprehension

Metaphor comprehension involves integrating meanings across domains to derive figurative interpretations. Prior work shows individual differences in inhibitory control affect L1 metaphor comprehension, moderated by metaphor conventionality and context. However, how inhibitory control influences L2 metaphor comprehension—where metaphor meanings are less salient and cross-linguistic mappings may be required—remains unclear. This study asks: (1) Do individuals with higher inhibitory control show processing advantages in L2 metaphor comprehension, and does this vary by metaphor conventionality (conventional vs. novel) and by the presence/absence of context? (2) Do individuals with higher inhibitory control derive greater benefits from supportive contexts than those with lower inhibitory control? The purpose is to clarify the flexible role of inhibitory control in L2 metaphor processing, extending the Structure Building Framework to bilingual figurative language. The study’s importance lies in explaining when and how inhibitory control supports or hinders L2 metaphor understanding depending on conventionality and contextual demands.

• Inhibitory control supports metaphor comprehension by suppressing irrelevant meanings and selecting context-appropriate interpretations (Structure Building Framework). Individuals with higher inhibitory control often outperform those with lower control on metaphor tasks, especially for novel metaphors that require controlled comparison, while conventional metaphors rely more on automatic categorization and retrieval of established senses.
• The career of metaphor hypothesis distinguishes novel (less familiar, requiring effortful mapping) from conventional metaphors (more familiar, category-based retrieval). Novel metaphors typically incur more processing cost than conventional ones.
• In L1 studies, higher executive control facilitates integrating contextual cues during metaphor processing, with high-control readers benefiting earlier and more efficiently from context than low-control readers.
• In L2, metaphorical meanings are less salient and comprehension is slower and more context-dependent. L2 metaphor processing can be less sensitive to conventionality and often relies on L1 knowledge, especially without context. Prior studies link executive control (including inhibitory control and working memory) to better L2 metaphor performance. Supportive contexts generally facilitate L2 metaphor comprehension, but benefits may be modulated by executive abilities. The specific contribution of inhibitory control across metaphor types and contexts in L2, however, has been understudied.

Design: Two experiments examined the role of inhibitory control in L2 metaphor comprehension as a function of metaphor conventionality (conventional vs. novel) and context (absent vs. supportive vs. literal). Inhibitory control was measured using a modified flanker task; participants were grouped into high inhibitory control (HIC) and low inhibitory control (LIC) based on inhibition scores (lower scores = higher inhibitory control).

Participants:

• Experiment 1: 87 Chinese-English bilingual undergraduates screened; 20 with mid-range inhibition scores excluded. Remaining 67 (34 HIC, 33 LIC; 47 females; mean age 19.5±1.4). Right-handed, normal/corrected vision, no neurological/psychiatric history. LexTALE scores did not differ between groups. After excluding 3 participants with <70% accuracy, final N=64 (equal HIC/LIC).
• Experiment 2: 80 screened; 14 mid-range excluded. 66 selected (33 HIC, 33 LIC; 45 females; mean age 19.8±1.85). Groups differed in inhibition scores but not in LexTALE. One LIC participant excluded for <80% accuracy, yielding N=64 (32 per group). Ethics: Approved by Xi’an Jiaotong University Health Science Center IRB (2022–1535). Written informed consent obtained.

Inhibitory control measure (modified flanker task):

• 144 trials (two blocks): 24 congruent, 24 incongruent, 24 neutral per block. Target arrow centrally presented with flanking arrows/boxes. Timing per trial: fixation 500 ms, blank 250 ms, stimulus up to 2000 ms, ISI standard. Inhibition score = mean RT(incongruent) − mean RT(congruent). Lower scores indicate greater inhibitory control. HIC range (Exp1): 14.94–51.88 (M=34.93±11.25); LIC range: 62.19–105.22 (M=75.19±11.36); groups differed p<0.001. Similar ranges in Exp2 (HIC: 15.05–49.89; LIC: 59.35–128.62), p<0.001.

Materials:

• Metaphor stimuli: 60 English nominal metaphors (X is Y): 30 conventional, 30 novel. Initially 80 were collected/modified from prior sources. Item norms with independent L2 English speakers: conventionality, comprehensibility, aptness (7-point scales); emotional valence, arousal, subjective frequency; sentence length and keyword letter counts assessed. Final selection ensured conventional vs. novel differed in conventionality (p<0.001) but matched on other dimensions.
• Experiment 1 fillers: 30 literal sentences and 90 anomalous sentences; total 180 trials per participant.
• Experiment 2 contexts: For each metaphor, two contexts created: supportive (figuratively biasing) and literal. Additional 60 literal sentence fillers each with supportive and literal contexts. A separate group rated target-context semantic relatedness (7-point scale); supportive contexts were more related than literal (t=4.43, p<0.001). Full materials available: https://osf.io/sf9n5/.

Procedures:

• Experiment 1 (no context): Semantic meaningfulness judgment. Each of 6 blocks had 30 trials (5 conventional, 5 novel, 5 literal, 15 anomalous). Trial timing: fixation 500 ms, blank 250 ms, sentence up to 5000 ms or response; ITI 800–1200 ms. Responses via Z/M keys; 12 practice trials.
• Experiment 2 (with context): Figurative meaning judgment to ensure context processing. Six blocks (180 trials): in each block, 10 conventional, 10 novel, 10 literal sentences, half preceded by supportive, half by literal contexts. Trial timing: fixation 500 ms; context 3000 ms; blank 250 ms; first half of target 1500 ms; blank 250 ms; second half up to 3000 ms or response. Upon second half onset, participants judged whether the whole sentence conveyed a figurative meaning (Z/M keys). 12 practice trials.

Data processing and statistical analysis:

• Experiment 1: Excluded participants with low accuracy (<70%). Removed incorrect trials (22.1%) and RTs beyond ±3 SD of participant mean (0.8% of correct trials). Mixed-effects linear models (lme4, lmerTest) on log RTs with fixed effects: group (HIC vs. LIC), type (conventional vs. novel), and interaction; random intercepts for subjects and items and random effect for participant by type. Logistic mixed-effects models for accuracy; best model included only type as fixed effect.
• Experiment 2: One participant excluded for <80% accuracy. Removed incorrect trials (6.4%) and RT outliers beyond ±3 SD (1.2% of correct trials). Mixed-effects linear models on log RTs with fixed effects: group, type, context (supportive vs. literal), and interactions; random intercepts for subjects and items and random effect for participant by type. Logistic mixed models for accuracy; best model included context as fixed effect. Backward model selection via step function; significance assessed via ANOVA chi-square and model coefficients.

Experiment 1 (no context):

• RTs: Significant interaction Group × Type (chi-square p=0.033). Pairwise: HIC slower than LIC for novel metaphors (p=0.023); no group difference for conventional metaphors (p=0.241). Main effect of Type: novel > conventional (p<0.001).
• ACC: Only main effect of Type (B=-1.90, SE=0.27, z=-6.94, p<0.001), with lower accuracy for novel than conventional metaphors; no group effect.
• Descriptive RTs (ms): HIC: conventional 1834±661, novel 2394±809; LIC: conventional 1731±712, novel 2051±759. ACCs: conventional ≈0.90 (both groups); novel HIC 0.67, LIC 0.65.

Experiment 2 (with context):

• RTs: Main effects of Type (B=0.06, SE=0.02, t=2.58, p=0.012) and Context (B=-0.02, SE=0.01, t=-2.53, p=0.011). Significant three-way interaction Group × Type × Context (B=-0.08, SE=0.04, t=-2.16, p=0.031). Pairwise: HIC responded faster to novel metaphors in supportive vs. literal contexts (p=0.013); LIC showed no such facilitation for novel metaphors. For conventional metaphors, LIC was faster in supportive vs. literal contexts (p=0.026); HIC showed no difference.
• ACC: Marginal effect of Context (B=0.19, SE=0.10, z=1.80, p=0.072), indicating a trend toward higher accuracy in supportive contexts; no effects of Group or Type.
• Descriptive RTs (ms) by condition: HIC—conventional supportive 880±354, conventional literal 878±326; novel supportive 918±359, novel literal 955±374. LIC—conventional supportive 904±346, conventional literal 954±401; novel supportive 996±415, novel literal 976±396.

Overall: In the absence of context, higher inhibitory control was associated with slower processing of L2 novel metaphors and no difference for conventional metaphors. With context, supportive context facilitated HIC participants’ processing of novel metaphors and LIC participants’ processing of conventional metaphors, evidencing a flexible, context- and conventionality-dependent role of inhibitory control.

The findings address the research questions by showing that inhibitory control influences L2 metaphor comprehension in a manner contingent on metaphor conventionality and contextual support. Without context, inhibitory control did not benefit conventional metaphors and was associated with slower processing of novel metaphors, suggesting that HIC individuals may engage in more deliberative, cross-linguistic comparison and integration when constructing interpretations for L2 novel metaphors. When supportive contexts were provided, HIC participants quickly integrated relevant cues and suppressed irrelevant information, gaining an advantage specifically for novel metaphors. Conversely, LIC participants benefited from supportive contexts for conventional metaphors, likely relying on salient, context-reinforced meanings and simpler concept matching. These patterns extend the Structure Building Framework to bilingual metaphor processing by demonstrating that the recruitment of inhibitory control is dynamic—facilitatory when multiple, contextually relevant cues must be integrated (novel metaphors with supportive context) but minimal or even costly in context-free novel cases in L2. The results also help reconcile mixed findings in L2 literature by highlighting the interplay of individual control abilities, metaphor type, and contextual demands.

This study demonstrates that the contribution of inhibitory control to L2 metaphor comprehension is flexible and jointly modulated by metaphor conventionality and context. In context-free settings, inhibitory control minimally affects conventional metaphor processing but is associated with slower responses to novel metaphors, consistent with more deliberative, resource-intensive integration. With supportive context, individuals with high inhibitory control show selective facilitation for novel metaphors, while those with low inhibitory control benefit more for conventional metaphors. These results extend the Structure Building Framework to L2 metaphors, emphasizing dynamic recruitment of inhibitory control depending on contextual demands and metaphor type. Future research should employ more naturalistic, syntactically varied metaphor stimuli and a broader battery of inhibitory control measures (e.g., Simon, Stroop) to improve generalizability and construct validity.

• Stimuli were limited to nominal metaphors with simple syntactic structures, which may constrain generalizability to naturalistic discourse.
• Inhibitory control was assessed using only a modified flanker task; convergent measures (e.g., Simon, Stroop) were not included.
• Participants were intermediate-proficiency Chinese-English bilingual undergraduates, which may limit generality across proficiency levels and language pairs.
• Accuracy and RT-based grouping and analyses may be sensitive to task demands and item characteristics despite careful norming.