{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WXFVOTBYCJP7YNXJIMV5YEOMCD","short_pith_number":"pith:WXFVOTBY","schema_version":"1.0","canonical_sha256":"b5cb574c38125ffc36e9432bdc11cc10d5855b1f787a9e1c0a193dacab276c87","source":{"kind":"arxiv","id":"2505.17407","version":1},"attestation_state":"computed","paper":{"title":"Language Matters: How Do Multilingual Input and Reasoning Paths Affect Large Reasoning Models?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Cheng-Kuang Wu, Chieh-Yen Lin, Hung-yi Lee, Yun-Nung Chen, Yu Ying Chiu, Zhi Rui Tam","submitted_at":"2025-05-23T02:46:18Z","abstract_excerpt":"Large reasoning models (LRMs) have demonstrated impressive performance across a range of reasoning tasks, yet little is known about their internal reasoning processes in multilingual settings. We begin with a critical question: {\\it In which language do these models reason when solving problems presented in different languages?} Our findings reveal that, despite multilingual training, LRMs tend to default to reasoning in high-resource languages (e.g., English) at test time, regardless of the input language. When constrained to reason in the same language as the input, model performance decline"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2505.17407","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-23T02:46:18Z","cross_cats_sorted":[],"title_canon_sha256":"d045bb87f6b3a13bf125e2c6332fdb4b25a3c56c2813146f92f6e9483ad16735","abstract_canon_sha256":"f7697fbb9dcc7ff740690b0e9381a6524b7bbdfa418a7bb1023de2642406827b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:25.677593Z","signature_b64":"aZG6jFQUEgXqA2NLqf0suJG/YCgq5v0nNhtx8BExOWdsg1QcpnyvdsREJ1lFQ33+9WqjFyGMH57QkaPS325TDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b5cb574c38125ffc36e9432bdc11cc10d5855b1f787a9e1c0a193dacab276c87","last_reissued_at":"2026-07-05T11:08:25.677097Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:25.677097Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Language Matters: How Do Multilingual Input and Reasoning Paths Affect Large Reasoning Models?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Cheng-Kuang Wu, Chieh-Yen Lin, Hung-yi Lee, Yun-Nung Chen, Yu Ying Chiu, Zhi Rui Tam","submitted_at":"2025-05-23T02:46:18Z","abstract_excerpt":"Large reasoning models (LRMs) have demonstrated impressive performance across a range of reasoning tasks, yet little is known about their internal reasoning processes in multilingual settings. We begin with a critical question: {\\it In which language do these models reason when solving problems presented in different languages?} Our findings reveal that, despite multilingual training, LRMs tend to default to reasoning in high-resource languages (e.g., English) at test time, regardless of the input language. When constrained to reason in the same language as the input, model performance decline"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17407","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2505.17407/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2505.17407","created_at":"2026-07-05T11:08:25.677160+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17407v1","created_at":"2026-07-05T11:08:25.677160+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17407","created_at":"2026-07-05T11:08:25.677160+00:00"},{"alias_kind":"pith_short_12","alias_value":"WXFVOTBYCJP7","created_at":"2026-07-05T11:08:25.677160+00:00"},{"alias_kind":"pith_short_16","alias_value":"WXFVOTBYCJP7YNXJ","created_at":"2026-07-05T11:08:25.677160+00:00"},{"alias_kind":"pith_short_8","alias_value":"WXFVOTBY","created_at":"2026-07-05T11:08:25.677160+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.27715","citing_title":"Beyond Input Understanding: Diagnosing Multilingual Mathematical Reasoning with Directed Acyclic Trace Graphs","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2510.27269","citing_title":"Why Do Multilingual Reasoning Gaps Emerge in Reasoning Language Models?","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16937","citing_title":"No One Fits All: From Fixed Prompting to Learned Routing in Multilingual LLMs","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD","json":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD.json","graph_json":"https://pith.science/api/pith-number/WXFVOTBYCJP7YNXJIMV5YEOMCD/graph.json","events_json":"https://pith.science/api/pith-number/WXFVOTBYCJP7YNXJIMV5YEOMCD/events.json","paper":"https://pith.science/paper/WXFVOTBY"},"agent_actions":{"view_html":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD","download_json":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD.json","view_paper":"https://pith.science/paper/WXFVOTBY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17407&json=true","fetch_graph":"https://pith.science/api/pith-number/WXFVOTBYCJP7YNXJIMV5YEOMCD/graph.json","fetch_events":"https://pith.science/api/pith-number/WXFVOTBYCJP7YNXJIMV5YEOMCD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD/action/storage_attestation","attest_author":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD/action/author_attestation","sign_citation":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD/action/citation_signature","submit_replication":"https://pith.science/pith/WXFVOTBYCJP7YNXJIMV5YEOMCD/action/replication_record"}},"created_at":"2026-07-05T11:08:25.677160+00:00","updated_at":"2026-07-05T11:08:25.677160+00:00"}