The present study investigates to what extent the effect of cross-linguistic differences on L2 idiom comprehension is modulated by the presence of a context. Sixty students of German as a foreign language (L1 French) completed a comprehension test consisting of metaphorical idioms in the L2 that differed from their L1 equivalents conceptually and formally and were presented with or without context. The results show that an increasing degree of conceptual and formal distance as well as the absence of context are generally associated with lower performance in the idiom comprehension test. However, the analysis of interactions shows that the presence of the context was especially supportive for conceptually different items, whereas the facilitative effect of formal similarity considerably diminished with increasing conceptual distance. The interplay of cross-linguistic differences and context… 497 (2002) also underscores the salience of literal meaning, describing two different phases that underlie the process of idiom comprehension in the L2: in the first phase, learners formulate hypotheses without using the context and thus rely on the linguistic mechanism. This process can be facilitated when the L1 and the L2 have similar expressions. In the second phase, the learners test the different hypotheses against the contextual information provided and progressively verify or reject them. In this vein, the Model of Dual Idiom Representation (Abel, 2003) postulates that L2 idiom comprehension is mainly affected by the frequency and decomposability of different idioms. Those idioms that are hardly decomposable and occur relatively frequently in the L2 input are more likely to develop a separate lexical entry, allowing direct access to their figurative meaning. The opposite is true for those idioms that have a high degree of compositionality and are rarely encountered in the L2 input, which often triggers an analysis of their components. A major drawback of both views (compositional vs. non-compositional) is the limited empirical evidence in the field of L2 idiom comprehension when compared to L1 idiom comprehension. Only a few psycholinguistic studies on L2 idiom comprehension have been conducted to date, but they are not conclusive about the explanatory potential of both views. On the one hand, some studies (Columbus, 2010; Conklin and Schmitt, 2008) showed that L2 learners processed literal expressions faster than figurative expressions, which is in line with the patterns found in native speakers (e.g. Durrant and Doherty, 2010). These findings seem to support the holistic retrieval of idioms postulated by non-compositional approaches (Gibbs, 1994, 1986; Bobrow and Bell, 1973). However, some other authors suggest that these results could be biased because the participants were in an immersion context and had a relatively high amount of exposure to an authentic and rich L2 input (Boers and Lindstromberg, 2012). On the other hand, there is also empirical evidence for the compositional view on idiom comprehension as well as the higher salience of literal meanings. For example,., a cross-modal priming study by Cieślicka (2006) showed that idiomatic expressions prime visual targets related to their literal interpretation, which in turn accounts for the prior activation of the literal meaning of the words before accessing the figurative meaning. In the same vein, Siyanova-Chanturia and Schmitt (2011) did not find any advantage to processing idioms over novel phrases by L2 learners, which contrasted with the pattern observed with native speakers in the same experiment (see also Cieślicka et al., 2014). The authors also found that L2 learners processed idioms faster when they were used literally than when used figuratively, suggesting that the literal meaning is also accessed when inferring the figurative meaning. Taken together, previous research supports the idea that both views complement each other in that they account for different L2 idiom comprehension conditions and explain how different factors affect this process. Whereas the findings supporting the non-compositional view suggest that increased language proficiency and the exposure to L2 in immersion contexts trigger a direct The interplay of cross-linguistic differences and context… 513 Kecskes, Istvan. 2000. A cognitive-pragmatic approach to situation-bound utterances. Journal of Pragmatics 32 (5). 605–625. Lakoff, George and Mark Johnson. 1980. Metaphors we live by. Chicago: The University of Chicago Press. Langacker, Ronald W. 1991. Foundations of cognitive grammar. Descriptive application . Volume 2. Stanford: Stanford University Press. Libben, Maya R. and Debra A. Titone. 2008. The multidetermined nature of idiom processing. Memory & Cognition 36. 1103–1121. Littlemore, Jeannette, Krenmayr, Tina, Turner, James and Sarah Turner. 2014. An investigation into metaphor use at different levels of second language writing. Applied Linguistics 32 (4). 208– 429. Littlemore, Jeannette and Graham Low. 2006. Metaphoric competence and communicative language ability. Applied Linguistics 27 (2). 268–294. Liontas, John I. 2002. Context and idiom understanding in second languages. EUROSLA Yearbook, 2. 155–185 Pizarro Pedraza, Andrea. 2013. Tabú y eufemismo en la ciudad de Madrid: estudio sociolíngüístico- cognitivo de los conceptos sexuales. PhD Thesis. Universidad Complutense de Madrid. Siyanova-Chanturia, Anna; Conklin, Kathy and Norbert Schmitt. 2011. Adding more fuel to the fire: An eye-tracking study of idiom processing by native and non-native speakers. Second Language Research 27. 1–22. Skoufaki, Sophia. 2008. Investigating the source of idiom transparency intuitions. Metaphor and Symbol 24. 20–41. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and cognition . Oxford: Blackwell. Sprenger, Simone A.; Levelt, Willem J.M. and Gerard Kempen. 2006. Lexical

English phraseological units (multi-word expressions such as idioms, collocations, proverbs, etc.) can be classified by integrating their syntactic structure and degree of semantic compositionality. This theoretical article proposes a structured taxonomy that cross-classifies phraseological units by formal syntactic pattern (e.g. verb–object combinations, adjective–noun collocations, binomial phrases, clausal proverbs) and by semantic transparency (ranging from fully non-compositional idioms to partially compositional idioms, conventional collocations, and pragmatically fixed formulas or pragmatemes). Drawing on phraseological research and Meaning–Text Theory, we distinguish categories like full idioms (holistic, non-literal meanings), semi-idioms (partially analyzable metaphoric expressions), collocations (compositional yet convention-bound word combinations), and pragmatemes (fixed expressions tied to specific pragmatic contexts). Illustrative examples are provided for each category alongside clear classification criteria. The taxonomy accounts for structural diversity (from fixed noun phrases to whole proverbs) while highlighting the continuum of compositionality. The discussion addresses implications for linguistic analysis, cross-linguistic comparison (with brief reference to Azerbaijani equivalents), and theoretical modeling of the lexicon. This integrative classification aims to enhance our understanding of phraseological units, benefiting lexicology, semantics, syntax, and phraseology research by offering a comprehensive framework for describing multi-word expressions in English.

Semantic priming typically facilitates lexical access; however, this facilitation may reverse into inhibition under certain interferences, such as embodied conflicts in Second Language (L2) processing. For adult L2 learners, automatic literal sensorimotor simulations may disrupt metaphorical integration, potentially inducing inhibitory reversal in conventional action metaphors [e.g., Chinese " "(chī kuī, literally "eat loss," figuratively "suffer the loss")]-a key gap in bilingual cognition. This study examines this reversal in L2 Chinese metaphor processing versus L1. Forty-three Vietnamese-speaking L2 learners of Chinese (HSK 5-6) and forty-seven first-language (L1) Chinese speakers completed a delayed-response semantic plausibility judgment task with morpheme-mediated semantic priming. Targets-literal, conventional metaphorical, and unrelated verb-object (VO) constructions-were each preceded by their identical verb morpheme (e.g., "" primes ""). Mixed-effects models revealed an opposite directional reversal in L2 learners: facilitation in literal versus unrelated baseline (shorter reaction times [RTs]; reduced errors) but inhibition in metaphorical versus unrelated baseline (elevated errors; nonsignificant RTs). In contrast to L2's reversal pattern, L1 Chinese speakers exhibited uniform dual inhibition across literal and metaphorical conditions (elevated errors; nonsignificant RTs), with a significant Group × Condition interaction. This study reveals an L2-specific reversal of priming in action metaphors (literal facilitation vs. metaphorical inhibition, primarily evident in error rates), originating from a dynamic mismatch between embodied simulations and semantic integration, a process potentially involving increased inhibitory control demands, while remaining consistent with broader processing costs at the behavioral level. These findings offer insights into the double-edged role of L2 embodiment in language processing -- helping Literal while hurting Metaphor Learners may over-apply literal meanings or experience L1 interference, mistakenly associating it with food-related phrases such as “吃饭很香” (to eat with relish) or produce awkward abstract usages. While this primarily stems from the embodied mismatch-lacking the entrenched L1-like link between the lexical concepts and the corresponding bodily-social experience-other contributing factors may include L1-L2 semantic-conceptual mismatches (e.g., differing metaphorical extensions of the literal “吃, eat” across languages), the holistic versus decompositional processing of idiomatic expressions, or challenges in chunk-based learning of formulaic sequences. Most research highlights general processing delays or ERP differences rather than a qualitative reversal in priming dynamics. This specific gap underscores the need for a targeted paradigm to isolate literal interference within syntactically transparent but semantically semi-opaque/idiomatic VO compounds. Therefore, this study aims to fill these interconnected gaps, by employing a morpheme-mediated semantic priming paradigm with delayed response. Four-panel data visualization compares behavioral performance in metaphorical language processing between two groups (L1 in blue, L2 in orange) across unrelated, metaphorical, and literal conditions. Panels show L2 higher mean reaction times and error rates, with strongest group differences in metaphorical conditions. Error bars indicate ninety-five percent confidence intervals. Overall, L1 speakers responded substantially faster ( M = 404 ms) than L2 learners ( M = 505 ms). For L2 learners, literal targets produced significantly faster RTs than unrelated targets ( b = −0.138, p < 0.001); in contrast, metaphorical targets did not differ significantly from unrelated targets ( b = −0.057, p = 0.095), although they were significantly slower than literal targets ( b = 0.082, p = 0.001). For L1 speakers, no significant RT differences were shown across any conditions: metaphorical vs. unrelated ( b = 0.051, p = 0.070), literal vs. unrelated ( b = 0.03, p = 0.332), and metaphorical vs. literal ( b = 0.021, p = 0.523). Source of conflict: dynamic mismatch between literal simulation and metaphorical integration The observed inhibitory reversal in L2 learners could plausibly arise from general semantic competition or semantic integration difficulty—common in L2 metaphor processing due to weaker lexical-semantic links, lower salience of figurative meanings, or cross-linguistic conceptual mismatch ( Chen et al., 2013 ; Miller et al., 2025 ). However, language is not a unimodal system of abstract symbols but a multimodal, embodied system in which semantic and conceptual representations are grounded in and constituted by sensorimotor simulations ( Zwaan, 2014 ). In literal trials [e.g.,“洗衣 “(xǐ yī, wash clothes)], this simulation aligns seamlessly with the compositional semantic representation, yielding processing facilitation: shorter RTs and lower error rates relative to unrelated baselines. Our data corroborate this, with L2 learners showing robust literal facilitation (vs. unrelated baseline: p < 0.001 in RTs; p = 0.043 in errors), suggesting that cross-linguistic bodily experiences (e.g., washing actions) transfer effectively to support basic sensorimotor alignment ( Vukovic and Shtyrov, 2014 ). L2 learners may command the declarative knowledge of metaphors such as “brainwash,” yet fail to automatize the integration of sensorimotor simulations and the abstract conceptual domain. This gap is rooted in distinct learning trajectories: L1 concepts emerge within vivid, multimodal immersion, whereas adult L2 is often acquired through explicit, decontextualized methods (word lists, translation, etc.). Thus, bilingual differences hinge not only on connection weights but also on the tightness of cross-modal binding and the real-time resolution of such bindings. The “reversal” effect is a signature of this adaptive bilingual cognition.