{"id":"https://openalex.org/W4416034302","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.1103","title":"When Models Reason in Your Language: Controlling Thinking Language Comes at the Cost of Accuracy","display_name":"When Models Reason in Your Language: Controlling Thinking Language Comes at the Cost of Accuracy","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034302","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.1103"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.1103","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1103","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1103.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.1103.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113261156","display_name":"Jirui Qi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jirui Qi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100369744","display_name":"Shan Chen","orcid":"https://orcid.org/0000-0003-0912-7254"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shan Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089895667","display_name":"Zidi Xiong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zidi Xiong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103024385","display_name":"Raquel Fern\u00e1ndez","orcid":"https://orcid.org/0000-0002-3889-6194"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Raquel Fern\u00e1ndez","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039369605","display_name":"Danielle S. Bitterman","orcid":"https://orcid.org/0000-0003-0345-2232"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Danielle Bitterman","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5019968969","display_name":"Arianna Bisazza","orcid":"https://orcid.org/0000-0003-1270-3048"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Arianna Bisazza","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"20279","last_page":"20296"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10656","display_name":"Child and Animal Learning Development","score":0.17350000143051147,"subfield":{"id":"https://openalex.org/subfields/3204","display_name":"Developmental and Educational Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10656","display_name":"Child and Animal Learning Development","score":0.17350000143051147,"subfield":{"id":"https://openalex.org/subfields/3204","display_name":"Developmental and Educational Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T12090","display_name":"Language and cultural evolution","score":0.07519999891519547,"subfield":{"id":"https://openalex.org/subfields/3316","display_name":"Cultural Studies"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10730","display_name":"Language Development and Disorders","score":0.0575999990105629,"subfield":{"id":"https://openalex.org/subfields/3204","display_name":"Developmental and Educational Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.28459998965263367},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.2786000072956085},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.2721000015735626},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.25859999656677246}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5295000076293945},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36070001125335693},{"id":"https://openalex.org/C111472728","wikidata":"https://www.wikidata.org/wiki/Q9471","display_name":"Epistemology","level":1,"score":0.3303000032901764},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.289900004863739},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.2856000065803528},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.28459998965263367},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2786000072956085},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.2721000015735626},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.25859999656677246},{"id":"https://openalex.org/C511192102","wikidata":"https://www.wikidata.org/wiki/Q5156948","display_name":"Comprehension","level":2,"score":0.241799995303154}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.1103","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1103","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1103.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},{"id":"pmh:oai:pure.rug.nl:openaire/95437341-b55e-48e5-ad2a-4b86f3de8985","is_oa":true,"landing_page_url":"https://research.rug.nl/en/publications/95437341-b55e-48e5-ad2a-4b86f3de8985","pdf_url":null,"source":{"id":"https://openalex.org/S4306400420","display_name":"University of Groningen research database (University of Groningen / Centre for Information Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I169381384","host_organization_name":"University of Groningen","host_organization_lineage":["https://openalex.org/I169381384"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":null,"raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.1103","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1103","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1103.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2388848714","display_name":null,"funder_award_id":"819455","funder_id":"https://openalex.org/F4320320300","funder_display_name":"European Commission"},{"id":"https://openalex.org/G4352045484","display_name":"Polyglot Machines: Human-like Learning of Morphologically Rich Languages","funder_award_id":"VI.Vidi.221C.009","funder_id":"https://openalex.org/F4320321800","funder_display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek"},{"id":"https://openalex.org/G6280615682","display_name":"Low Resource Chat-based Conversational Intelligence","funder_award_id":"NWA.1389.20.183","funder_id":"https://openalex.org/F4320321800","funder_display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek"},{"id":"https://openalex.org/G8956110671","display_name":null,"funder_award_id":"U54CA274516","funder_id":"https://openalex.org/F4320332161","funder_display_name":"National Institutes of Health"}],"funders":[{"id":"https://openalex.org/F4320320300","display_name":"European Commission","ror":"https://ror.org/00k4n6c32"},{"id":"https://openalex.org/F4320320933","display_name":"Rijksuniversiteit Groningen","ror":"https://ror.org/012p63287"},{"id":"https://openalex.org/F4320321800","display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek","ror":"https://ror.org/04jsz6e67"},{"id":"https://openalex.org/F4320332161","display_name":"National Institutes of Health","ror":"https://ror.org/01cwqze88"}],"has_content":{"pdf":true,"grobid_xml":false},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034302.pdf"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"Large":[1],"Reasoning":[2],"Models":[3],"(LRMs)":[4],"with":[5,136],"thinking":[6,45,97],"traces":[7],"have":[8],"shown":[9],"strong":[10],"performance":[11],"on":[12,67],"English":[13,81],"reasoning":[14,85,156],"tasks.":[15],"However,":[16],"the":[17,44,74,94,106,153],"extent":[18],"to":[19,80,103],"which":[20],"LRMs":[21,66,160],"can":[22,140],"think":[23],"in":[24,53,86,93,98,105],"other":[25,87],"languages":[26],"is":[27,31,147],"less":[28],"studied.":[29],"This":[30,116],"as":[32,34],"important":[33,127],"answer":[35,123],"accuracy":[36,146],"for":[37,48,164],"real-world":[38],"applications":[39],"since":[40],"users":[41],"may":[42],"find":[43],"trace":[46],"useful":[47],"oversight":[49],"only":[50],"if":[51],"expressed":[52],"their":[54],"languages.":[55,100],"In":[56],"this":[57,142],"work,":[58],"we":[59],"comprehensively":[60],"evaluate":[61],"two":[62],"leading":[63],"families":[64],"of":[65,96,158],"our":[68],"established":[69],"benchmark":[70],"XReasoning.":[71],"Surprisingly,":[72],"even":[73,135],"most":[75],"advanced":[76],"models":[77,102],"often":[78],"revert":[79],"or":[82],"produce":[83],"fragmented":[84],"languages,":[88],"revealing":[89],"a":[90],"substantial":[91],"gap":[92],"capability":[95],"non-English":[99],"Promoting":[101],"reason":[104],"user\u2019s":[107],"language":[108,143],"via":[109],"prompt":[110],"hacking":[111],"enhances":[112],"readability":[113],"and":[114,161],"oversight.":[115],"could":[117],"gain":[118],"user":[119],"trust,":[120],"but":[121],"reduces":[122],"accuracy,":[124],"exposing":[125],"an":[126],"trade-off.":[128],"We":[129],"further":[130],"demonstrate":[131],"that":[132],"targeted":[133],"post-training,":[134],"just":[137],"100":[138],"instances,":[139],"mitigate":[141],"mismatch,":[144],"although":[145],"still":[148],"degraded.":[149],"Our":[150],"results":[151],"reveal":[152],"limited":[154],"multilingual":[155],"capabilities":[157],"current":[159],"suggest":[162],"directions":[163],"future":[165],"research.":[166]},"counts_by_year":[],"updated_date":"2026-07-20T07:56:41.581041","created_date":"2025-11-08T00:00:00"}