{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":26,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":26,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"ce48bc7d526f","filters":{"venue":"Linguistics Vanguard"}},"results":[{"id":"W2892039495","doi":"10.1515/lingvan-2017-0027","title":"The role of predictability in shaping phonological patterns","year":2018,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Phonetics and Phonology Research","field":"Psychology","cited_by":108,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Predictability; Context (archaeology); Computer science; Set (abstract data type); Comprehension; Reduction (mathematics); Cognitive psychology; Linguistics; Natural language processing; Psychology; Mathematics; Statistics; History","authors":[{"name":"Kathleen Currie Hall","is_ca":true},{"name":"Elizabeth Hume","is_ca":false},{"name":"T. Florian Jaeger","is_ca":false},{"name":"Andrew Wedel","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04052652632450414,"gpt":0.3620769572044678,"spread":0.3215504308799637,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001284488,0.0001831532,0.0002340887,0.0008858673,0.0003438132,0.002199627,0.0003783368,0.0004732226,0.002031639],"category_scores_gemma":[0.00922686,0.0003725785,0.0002381715,0.0005374508,0.001863825,0.001922655,0.0009356238,0.0007729818,0.0003507279],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000387519,"about_ca_system_score_gemma":0.0003911606,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001561466,"about_ca_topic_score_gemma":0.001771337,"domain_scores_codex":[0.9993755,0.0002264437,0.00004316968,0.0001414466,0.0001602548,0.0000532648],"domain_scores_gemma":[0.9953443,0.002978967,0.0008126888,0.0003347783,0.000411189,0.000118093],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0006844066,0.0001414099,0.302263,0.001122441,0.0004758549,0.0015551,0.006323581,0.04033889,0.1393499,0.1795911,0.002660461,0.3254938],"study_design_scores_gemma":[0.00003441499,0.0002437917,0.4722064,0.0002688697,0.0001826539,0.0008648451,0.001547804,0.05999786,0.01671831,0.4380307,0.009697301,0.0002070863],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8617993,0.003009979,0.1056826,0.001194054,0.00009127535,0.00003135933,0.0002755079,0.0003884231,0.02752764],"genre_scores_gemma":[0.9966852,0.000303681,0.00266054,0.00003737034,0.00001681795,0.000004068521,0.00002049871,0.00003301855,0.0002387533],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.002199627,"threshold_uncertainty_score":0.006796479,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2885411231","doi":"10.1515/lingvan-2017-0018","title":"Practice makes perfect: the consequences of lexical proficiency for articulation","year":2018,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Phonetics and Phonology Research","field":"Psychology","cited_by":64,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"Deutsche Forschungsgemeinschaft","keywords":"Coarticulation; Articulation (sociology); Speech recognition; Linguistics; Psychology; Computer science; Speech production; Cognitive psychology; Constraint (computer-aided design); Mathematics; Vowel","authors":[{"name":"Fabian Tomaschek","is_ca":false},{"name":"Benjamin V. Tucker","is_ca":true},{"name":"Matteo Fasiolo","is_ca":false},{"name":"R. Harald Baayen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06326567726277735,"gpt":0.4367760378886497,"spread":0.3735103606258723,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00141012,0.0002049926,0.0002778518,0.0003918047,0.0003011298,0.001678712,0.0003309472,0.0006752895,0.004248402],"category_scores_gemma":[0.02434888,0.0002722763,0.0001444799,0.0003206114,0.001164825,0.0009481388,0.001180072,0.0004978516,0.0003058591],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002736296,"about_ca_system_score_gemma":0.0002534099,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009939219,"about_ca_topic_score_gemma":0.001008939,"domain_scores_codex":[0.9988266,0.0004111267,0.0001022606,0.000301595,0.0002207741,0.0001376879],"domain_scores_gemma":[0.9823869,0.01294012,0.002342171,0.001020592,0.000516252,0.0007939508],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.003022701,0.001040118,0.5502158,0.0004880426,0.0002466077,0.002204183,0.01098071,0.006902239,0.3290694,0.00458603,0.0003651792,0.09087897],"study_design_scores_gemma":[0.00003872282,0.0009008631,0.9780508,0.00005384861,0.00004980495,0.0005027688,0.002964419,0.004140926,0.008062814,0.00457729,0.00061959,0.00003818318],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9974902,0.00005841993,0.0009045228,0.00005127211,0.000002706712,0.000005565022,0.000030804,0.000008554237,0.001447921],"genre_scores_gemma":[0.9995905,0.00001387368,0.0002722753,0.000009561175,0.000001669409,0.000003543811,0.00001167788,0.000005187247,0.00009171497],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004248402,"threshold_uncertainty_score":0.01421237,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2800761643","doi":"10.1515/lingvan-2016-0081","title":"The syntax-prosody interface: current theoretical approaches and outstanding questions","year":2018,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Phonetics and Phonology Research","field":"Psychology","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Prosody; Syntax; Interface (matter); Linguistics; Computer science; Situated; Field (mathematics); Natural language processing; Recursion (computer science); Current (fluid); Artificial intelligence; Programming language; Physics; Mathematics; Speech recognition; Philosophy","authors":[{"name":"Emily Elfner","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06771065085330953,"gpt":0.4011014568515419,"spread":0.3333908059982323,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01406227,0.0009274486,0.001674319,0.005287951,0.002315171,0.0177825,0.004182697,0.005303512,0.007688561],"category_scores_gemma":[0.01114325,0.00103951,0.000657611,0.004266322,0.03392844,0.02875087,0.005484027,0.00848369,0.001667951],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00382577,"about_ca_system_score_gemma":0.003857618,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001943418,"about_ca_topic_score_gemma":0.001228569,"domain_scores_codex":[0.9956064,0.002185068,0.0002673482,0.000821354,0.0008127448,0.0003070566],"domain_scores_gemma":[0.9820325,0.01351105,0.0009723061,0.001260797,0.00166034,0.0005628674],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00003716743,0.00004574411,0.0004685378,0.0003800627,0.00001722189,0.00003591376,0.001284822,0.0005088118,0.0001618617,0.9628197,0.001146291,0.03309387],"study_design_scores_gemma":[0.000006431041,0.00002156461,0.0004074614,0.0004806619,0.000009926594,0.00004982224,0.002000719,0.002006172,0.0001347846,0.9833875,0.0114745,0.00002050031],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.04454263,0.5300721,0.1164269,0.184144,0.001971286,0.00006319904,0.0002459377,0.000413975,0.12212],"genre_scores_gemma":[0.7502141,0.1980903,0.03274493,0.00777447,0.006002003,0.0002200439,0.0001972444,0.0001851173,0.004571835],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.0177825,"threshold_uncertainty_score":0.07436931,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2609372572","doi":"10.1515/lingvan-2015-0032","title":"Is <i>like</i> like <i>like</i>?: Evaluating the same variant across multiple variables","year":2017,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":35,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Salient; Meaning (existential); Variation (astronomy); Perception; Linguistics; Psychology; Variable (mathematics); Verb; Perspective (graphical); Transitive relation; Cognitive psychology; Social psychology; Mathematics; Computer science","authors":[{"name":"Ruth Maddeaux","is_ca":true},{"name":"Aaron J. Dinkin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07200595367626476,"gpt":0.3984130001953998,"spread":0.326407046519135,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002678308,0.0001822884,0.0002660666,0.0007649915,0.0006143376,0.002727747,0.0003938986,0.0005687171,0.002332554],"category_scores_gemma":[0.01285962,0.0001387974,0.0002007506,0.0006754951,0.002171103,0.002439566,0.0010744,0.0006633141,0.0002539183],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000505243,"about_ca_system_score_gemma":0.0002240452,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001911373,"about_ca_topic_score_gemma":0.002032405,"domain_scores_codex":[0.9983681,0.0008230313,0.00008300823,0.000292364,0.0003323044,0.0001013007],"domain_scores_gemma":[0.9902787,0.005493017,0.001546205,0.00106618,0.00127708,0.0003388065],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00238597,0.0002720101,0.4607697,0.0005634681,0.0002860982,0.001287914,0.1548852,0.001811861,0.2148962,0.0506091,0.0009541202,0.1112784],"study_design_scores_gemma":[0.00004018796,0.0009296733,0.8103172,0.0001320823,0.0002460635,0.001683649,0.1133473,0.009558477,0.02622245,0.02692417,0.01033497,0.0002638429],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9847404,0.00006772482,0.003608308,0.00007687163,0.0000118122,0.00001282502,0.00004159318,0.00002013142,0.01142019],"genre_scores_gemma":[0.9990863,0.00001501671,0.0006788966,0.0000131687,0.000002729931,0.000003717943,0.00002543238,0.00001091405,0.0001637471],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.002727747,"threshold_uncertainty_score":0.01416439,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2030992269","doi":"10.1515/lingvan-2014-1008","title":"Language structure and social agency: Confirming polar questions in conversation","year":2015,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Language, Discourse, Communication Strategies","field":"Arts and Humanities","cited_by":25,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Semiotics; Linguistics; Conversation; Agency (philosophy); Repetition (rhetorical device); Function (biology); Conversation analysis; Selection (genetic algorithm); Computer science; Epistemology; Psychology; Artificial intelligence; Philosophy","authors":[{"name":"N. J. Enfield","is_ca":false},{"name":"Jack Sidnell","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04871211242979754,"gpt":0.3079362462959112,"spread":0.2592241338661136,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01269796,0.000446195,0.0004167997,0.001909181,0.004834468,0.00604977,0.0009471329,0.002544777,0.004189197],"category_scores_gemma":[0.02978906,0.0005085238,0.000525175,0.00102747,0.01544965,0.007840525,0.005148062,0.001886327,0.0005889325],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002315204,"about_ca_system_score_gemma":0.001230801,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003940444,"about_ca_topic_score_gemma":0.002563887,"domain_scores_codex":[0.9687845,0.0267019,0.0006326896,0.00130387,0.001578019,0.0009991266],"domain_scores_gemma":[0.963883,0.02672344,0.003734477,0.002418288,0.002665282,0.0005754526],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0003796806,0.0000947983,0.01293932,0.0002842677,0.00005933673,0.002352497,0.4928533,0.001468901,0.02148188,0.4418913,0.0008490581,0.02534559],"study_design_scores_gemma":[0.0001288683,0.0005868487,0.02737753,0.0005504927,0.0001810232,0.002667222,0.5106375,0.01552681,0.02670419,0.3160465,0.09922816,0.0003648898],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.824407,0.0003296607,0.0672528,0.003765763,0.00005486226,0.0001315314,0.0000565165,0.000239031,0.1037629],"genre_scores_gemma":[0.9972954,0.00002304283,0.001909012,0.00006521786,0.000008028803,0.00001511071,0.000007624998,0.00001550216,0.0006609473],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01269796,"threshold_uncertainty_score":0.06715405,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2335647126","doi":"10.1515/lingvan-2015-0010","title":"An evaluation of noise on LPC-based vowel formant estimates: Implications for sociolinguistic data collection","year":2016,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Memorial University of Newfoundland","funders":"","keywords":"Formant; Spurious relationship; Noise (video); Vowel; Computer science; Speech recognition; Data collection; SIGNAL (programming language); Interference (communication); Acoustics; Telecommunications; Artificial intelligence; Mathematics; Statistics; Physics; Machine learning","authors":[{"name":"Paul De Decker","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1564448628560941,"gpt":0.4492947400870724,"spread":0.2928498772309783,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01402082,0.0007251253,0.0007039771,0.0009049243,0.0009377183,0.001527016,0.0008952586,0.0008698169,0.001399201],"category_scores_gemma":[0.07671409,0.000429444,0.0004744048,0.0008268086,0.001554499,0.0009606233,0.001264955,0.0004924449,0.0004373932],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005886326,"about_ca_system_score_gemma":0.0004547104,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003520458,"about_ca_topic_score_gemma":0.004535027,"domain_scores_codex":[0.9843269,0.01036139,0.0007663144,0.001124076,0.003198653,0.0002226442],"domain_scores_gemma":[0.8776476,0.1034761,0.003212393,0.00498201,0.01000272,0.000679141],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.02456686,0.002036972,0.1192238,0.001523605,0.0004958719,0.0006734834,0.01343912,0.02086169,0.543291,0.001503166,0.0007393616,0.2716451],"study_design_scores_gemma":[0.0003472313,0.016724,0.6484089,0.0002007217,0.0005627674,0.0007719486,0.004747114,0.03824091,0.2848621,0.001569731,0.003134183,0.0004304456],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.9480391,0.0001984454,0.04885719,0.0001017497,0.00005075546,0.0003309858,0.0002831137,0.000242033,0.001896674],"genre_scores_gemma":[0.947252,0.0001257309,0.05115393,0.00006947851,0.00004278504,0.00042039,0.000269545,0.0001530656,0.0005131041],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.01402082,"threshold_uncertainty_score":0.07415003,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2565228806","doi":"10.1515/lingvan-2015-0012","title":"Extending ELAN into variationist sociolinguistics","year":2015,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Coding (social sciences); Transcription (linguistics); Linguistics; Natural language processing; Statistics; Mathematics","authors":[{"name":"Naomi Nagy","is_ca":true},{"name":"Miriam Meyerhoff","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05606557689429376,"gpt":0.3702220372922769,"spread":0.3141564603979831,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01260467,0.0009592381,0.0006779068,0.004715949,0.00184903,0.004736464,0.002148536,0.0005693791,0.01685469],"category_scores_gemma":[0.021764,0.0006948814,0.0005856524,0.003165684,0.00388926,0.005703627,0.007214288,0.002874062,0.003343151],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002505693,"about_ca_system_score_gemma":0.001873147,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002561207,"about_ca_topic_score_gemma":0.005222761,"domain_scores_codex":[0.9865116,0.01048305,0.0005425039,0.001136941,0.001146081,0.0001799353],"domain_scores_gemma":[0.9768432,0.0168169,0.0006321949,0.003671423,0.001811886,0.00022431],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001289889,0.0001129668,0.001674605,0.0006859301,0.0000468891,0.0004517836,0.04600328,0.003294732,0.003187218,0.4641691,0.01510969,0.4651348],"study_design_scores_gemma":[0.00003664935,0.00006349389,0.00362891,0.0008081776,0.00002797655,0.0005295522,0.008195458,0.03242769,0.003753124,0.5062861,0.4441368,0.0001059339],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01092568,0.0002195699,0.9315479,0.001105472,0.000131526,0.000309762,0.001145015,0.003050483,0.05156449],"genre_scores_gemma":[0.1274349,0.0002757631,0.8604953,0.0003412369,0.00008426132,0.001030563,0.001456723,0.001511881,0.007369409],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01685469,"threshold_uncertainty_score":0.06666064,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4306249762","doi":"10.1515/lingvan-2021-0122","title":"Phonetic change over the career: a case study","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"European Commission","keywords":"Flexibility (engineering); Perspective (graphical); Psychology; Linguistics; Longitudinal study; Feature (linguistics); Computer science; Mathematics","authors":[{"name":"Josiane Riverin-Coutlée","is_ca":false},{"name":"Jonathan Harrington","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07789135288875891,"gpt":0.3521037197557527,"spread":0.2742123668669938,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008649828,0.0004568843,0.0003727043,0.001593175,0.00439943,0.00124658,0.0006577007,0.002010349,0.002100956],"category_scores_gemma":[0.002862653,0.0002690717,0.0004485721,0.001278946,0.001621295,0.000602986,0.001321012,0.001324542,0.0003210249],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002098342,"about_ca_system_score_gemma":0.001152605,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03222817,"about_ca_topic_score_gemma":0.06346511,"domain_scores_codex":[0.999178,0.0002394079,0.00004414362,0.0001336233,0.0001484127,0.000256305],"domain_scores_gemma":[0.9987465,0.0004488353,0.0001964756,0.00009020563,0.0001627687,0.0003552705],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"case_report","study_design_gemma":"qualitative","study_design_scores_codex":[0.0002128309,0.0005063215,0.1859753,0.0001445975,0.00003259049,0.5989397,0.1694021,0.0003267354,0.003925375,0.002027103,0.001693242,0.03681409],"study_design_scores_gemma":[0.00001760401,0.0005038285,0.1397706,0.0002095531,0.00006564729,0.6490694,0.1831943,0.001365766,0.00306732,0.000974197,0.02164564,0.0001161486],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9963232,0.0003647823,0.0004881767,0.0006037385,0.00003877679,0.00002076425,0.0000635973,0.000008591819,0.002088351],"genre_scores_gemma":[0.9961911,0.0004447602,0.0004921587,0.0001600409,0.00004005347,0.00001229432,0.00003164824,0.000008816756,0.002619049],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03222817,"threshold_uncertainty_score":0.06408119,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2590027471","doi":"10.1515/lingvan-2016-0025","title":"Individual differences in second language speech perception across tasks and contrasts: The case of English vowel contrasts by Korean learners","year":2017,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Phonetics and Phonology Research","field":"Psychology","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Santé, Sciences Biologiques et Chimie du Vivant","keywords":"Psychology; Perception; Vowel; Formant; Weighting; Task (project management); Contrast (vision); Duration (music); Cognitive psychology; Speech perception; Natural (archaeology); Two-alternative forced choice; American English; Linguistics; Speech recognition; Computer science; Artificial intelligence","authors":[{"name":"Donghyun Kim","is_ca":true},{"name":"Meghan Clayards","is_ca":true},{"name":"Heather Goad","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02822879181188652,"gpt":0.3582191485080133,"spread":0.3299903566961268,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007958353,0.0003274032,0.000287671,0.0004970763,0.0002241574,0.0009688326,0.000174243,0.0003552296,0.001430113],"category_scores_gemma":[0.003938208,0.0002470719,0.0002347217,0.0002170927,0.0004595157,0.0005382895,0.0005949399,0.0003820305,0.0003407662],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001373931,"about_ca_system_score_gemma":0.0001353523,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001343096,"about_ca_topic_score_gemma":0.001648905,"domain_scores_codex":[0.999619,0.00007804642,0.00004657996,0.0001040452,0.000103032,0.00004924533],"domain_scores_gemma":[0.9969224,0.001327125,0.0006595072,0.000307004,0.0004582821,0.0003256572],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001176234,0.0004520601,0.7381172,0.0001033961,0.0001899439,0.0007135423,0.01727618,0.0005957854,0.2191431,0.0001684756,0.0001701698,0.02189378],"study_design_scores_gemma":[0.00001560341,0.0006729399,0.9742495,0.00001137431,0.0000423343,0.000923874,0.006114316,0.001109747,0.0162203,0.0001624797,0.0004347609,0.00004280351],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9997955,0.00001132112,0.00006210061,0.000002584184,6.913876e-7,0.000001671453,0.000009837932,0.000001569475,0.0001146615],"genre_scores_gemma":[0.9996622,0.00001307942,0.0001090739,0.000005071752,6.881893e-7,0.000003979003,0.00002112323,0.000002949942,0.000181903],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.001430113,"threshold_uncertainty_score":0.004784226,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4214853268","doi":"10.1515/lingvan-2021-0057","title":"Disruptions due to COVID-19: using mixed methods to identify factors influencing language maintenance and shift","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Multilingual Education and Policy","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Context (archaeology); Social distance; Recreation; Language shift; Perception; Psychology; Distancing; Affect (linguistics); Coronavirus disease 2019 (COVID-19); Identity (music); Heritage language; Social psychology; Sociology; Pedagogy; Linguistics; Geography; Political science; Medicine; Communication","authors":[{"name":"Maya Ravindranath Abtahian","is_ca":false},{"name":"Naomi Nagy","is_ca":true},{"name":"Katharina Pabst","is_ca":true},{"name":"Vidhya Elango","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1032113126007449,"gpt":0.5455870256524398,"spread":0.4423757130516949,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04812368,0.0005157096,0.001045524,0.003954581,0.003574242,0.004285275,0.001975819,0.001239364,0.002982811],"category_scores_gemma":[0.06878462,0.0007393213,0.001230658,0.00432187,0.002717988,0.002584277,0.005412763,0.001646514,0.0002699116],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00448333,"about_ca_system_score_gemma":0.005184607,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.015003,"about_ca_topic_score_gemma":0.02547955,"domain_scores_codex":[0.9606546,0.02836856,0.003603655,0.002158715,0.003491645,0.0017228],"domain_scores_gemma":[0.9203656,0.05538538,0.0119313,0.004882918,0.006039014,0.001395847],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"qualitative","study_design_scores_codex":[0.000748281,0.001164556,0.4515903,0.002529194,0.0007243437,0.0007960569,0.4510954,0.0006230854,0.002002006,0.004385925,0.001651747,0.08268916],"study_design_scores_gemma":[0.00009096754,0.001949473,0.4100026,0.001973638,0.0003459173,0.0002748849,0.564123,0.003467235,0.003669666,0.003848838,0.0100571,0.0001965673],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9829948,0.0007669582,0.009943902,0.0004666184,0.00004483615,0.003140581,0.0006080844,0.00002789292,0.002006283],"genre_scores_gemma":[0.9768574,0.0004236291,0.01259017,0.0002993419,0.00002189221,0.008448438,0.0003238504,0.00002506536,0.001010239],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04812368,"threshold_uncertainty_score":0.2545055,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4312200521","doi":"10.1515/lingvan-2022-0017","title":"The Red Hen Anonymizer and the Red Hen Protocol for de-identifying audiovisual recordings","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Hearing Impairment and Communication","field":"Psychology","cited_by":4,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"Fundación Séneca; Google; Alexander von Humboldt-Stiftung","keywords":"JSON; Computer science; Software; Protocol (science); Identification (biology); Face (sociological concept); Multimedia; Human–computer interaction; Speech recognition; World Wide Web; Programming language; Linguistics","authors":[{"name":"Yash Khasbage","is_ca":false},{"name":"Daniel Alcaraz Carrión","is_ca":false},{"name":"Jennifer Hinnell","is_ca":true},{"name":"Frankie Robertson","is_ca":false},{"name":"Karan Singla","is_ca":false},{"name":"Peter Uhrig","is_ca":false},{"name":"Mark Turner","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06178125760216429,"gpt":0.4070607818337643,"spread":0.3452795242316,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03626467,0.001025665,0.0009568845,0.003777514,0.002930224,0.004091033,0.002221333,0.002478948,0.1470459],"category_scores_gemma":[0.07201785,0.001454204,0.0006805212,0.002229728,0.002914648,0.005624579,0.00638518,0.00450379,0.0699674],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001598789,"about_ca_system_score_gemma":0.005599926,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001373503,"about_ca_topic_score_gemma":0.001791032,"domain_scores_codex":[0.9683779,0.01649209,0.004859952,0.002312932,0.006569215,0.00138784],"domain_scores_gemma":[0.9185116,0.02525934,0.003995008,0.03584022,0.0149955,0.001398449],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.002068595,0.0002727294,0.001508769,0.001080312,0.00005652736,0.001385968,0.005955906,0.001550263,0.01954908,0.13411,0.6670112,0.1654506],"study_design_scores_gemma":[0.0001509077,0.0001286727,0.001910154,0.0006810707,0.0000254298,0.0005626461,0.001032494,0.003767979,0.02145889,0.02754174,0.9425935,0.0001464661],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01073794,0.0003570587,0.7666093,0.006454321,0.002933656,0.02033312,0.03338702,0.03223015,0.1269575],"genre_scores_gemma":[0.1021547,0.001182365,0.5088083,0.006679698,0.001647906,0.07537706,0.05200663,0.01701937,0.235124],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.1470459,"threshold_uncertainty_score":0.4919176,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4210262403","doi":"10.1515/lingvan-2020-0013","title":"A socially anchored approach to spatial language in Kalaallisut","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Categorization, perception, and language","field":"Psychology","cited_by":3,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Sociocultural evolution; Affect (linguistics); Encoding (memory); Space (punctuation); Arctic; Geography; Sociology; Frame (networking); Linguistics; Psychology; Computer science; Communication; Ecology; Cognitive psychology; Anthropology; Philosophy","authors":[{"name":"Hilary McMahan","is_ca":false},{"name":"Lenore A. Grenoble","is_ca":false},{"name":"Alliaq Kleist Petrussen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01596708737772196,"gpt":0.3064200608114498,"spread":0.2904529734337278,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001008755,0.0003075272,0.0002236146,0.001352261,0.006505511,0.006379105,0.0007300528,0.0005891633,0.002447919],"category_scores_gemma":[0.001328277,0.0001744483,0.0001513265,0.0008600804,0.01479242,0.00320598,0.004467801,0.001385419,0.0001767364],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.006219429,"about_ca_system_score_gemma":0.002196528,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05485366,"about_ca_topic_score_gemma":0.08709829,"domain_scores_codex":[0.9986591,0.0007231751,0.00004405674,0.0002043331,0.0001642752,0.0002049553],"domain_scores_gemma":[0.9993324,0.0002921704,0.0001307785,0.00006021605,0.00009555667,0.00008885455],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0000370047,0.00004320334,0.007477561,0.00005000447,0.000009146718,0.0007482369,0.6692755,0.0003010242,0.003182306,0.3066818,0.0004075444,0.01178676],"study_design_scores_gemma":[0.00001387393,0.00008129758,0.02841803,0.0001965047,0.00003818868,0.001293953,0.7313076,0.004397591,0.00257135,0.1338712,0.09772044,0.00009004882],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8111339,0.000763728,0.01210431,0.002190831,0.00006136624,0.00002177826,0.00002804304,0.00005021155,0.1736458],"genre_scores_gemma":[0.9976426,0.00006184993,0.00081755,0.00003148209,0.000003115669,0.000003902398,0.000005106596,0.000006938661,0.001427488],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.05485366,"threshold_uncertainty_score":0.1090688,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4285586648","doi":"10.1515/lingvan-2021-0048","title":"Cross-dialectal synchronic variation of a diachronic conditioned merger in Tlingit","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Phonetics and Phonology Research","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"McGill University; University of British Columbia","funders":"","keywords":"Variation (astronomy); Sound change; Perception; Linguistics; Indigenous; Set (abstract data type); History; Computer science; Psychology; Biology","authors":[{"name":"Amanda Cardoso","is_ca":true},{"name":"James Crippen","is_ca":true},{"name":"Gloria Mellesmoen","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02094465324947253,"gpt":0.3604735833782272,"spread":0.3395289301287547,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004414982,0.0001234614,0.0001691125,0.0007776331,0.0007430455,0.0008455729,0.0001856262,0.0001998249,0.003309136],"category_scores_gemma":[0.001700283,0.0001283718,0.00006804236,0.0007719881,0.001118321,0.0003395244,0.001124671,0.0004157743,0.0002679124],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007418027,"about_ca_system_score_gemma":0.0003595728,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01729225,"about_ca_topic_score_gemma":0.04170005,"domain_scores_codex":[0.9996915,0.00004627499,0.00002998799,0.0001117758,0.00006101609,0.00005943957],"domain_scores_gemma":[0.9988853,0.0003432082,0.0002073605,0.00011527,0.0003334781,0.0001153642],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0008410374,0.0001245084,0.4983607,0.000144331,0.00007768875,0.001451747,0.03913398,0.0005537784,0.4203755,0.002442125,0.0003385801,0.0361561],"study_design_scores_gemma":[0.000002501223,0.00006436275,0.9865348,0.00001103154,0.00001564913,0.0003638842,0.006714986,0.0002509335,0.004751037,0.000277029,0.0009979574,0.0000158585],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9977282,0.00002235853,0.0002553938,0.000007324522,0.000003499248,0.000002957981,0.00004258382,0.000005705312,0.001932121],"genre_scores_gemma":[0.9996073,0.000007659851,0.0001100126,0.00000429204,9.353034e-7,0.000002471257,0.00003017155,0.000006426135,0.0002307944],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01729225,"threshold_uncertainty_score":0.03438324,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4210791957","doi":"10.1515/lingvan-2020-0016","title":"Conflation of spatial reference frames in deaf community sign languages","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Hearing Impairment and Communication","field":"Psychology","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Modality (human–computer interaction); Conflation; Affordance; Sign language; Modalities; Sign (mathematics); American Sign Language; Frame of reference; Frame (networking); Computer science; Linguistics; Manually coded language; Feature (linguistics); Sociolinguistics of sign languages; Psychology; Artificial intelligence; Sociology; Human–computer interaction; Mathematics","authors":[{"name":"Oksana Tkachman","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0589559648813121,"gpt":0.3698895156348347,"spread":0.3109335507535226,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002006729,0.0002784881,0.0004581679,0.003358434,0.000497658,0.002997213,0.0006529002,0.0005490598,0.002258803],"category_scores_gemma":[0.005313483,0.0001673129,0.000250769,0.002721981,0.003618923,0.002696524,0.001563812,0.0005190307,0.0002525445],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009424539,"about_ca_system_score_gemma":0.001221209,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00362506,"about_ca_topic_score_gemma":0.00318357,"domain_scores_codex":[0.9986817,0.0005238194,0.0001862929,0.0001972976,0.0003303621,0.00008043925],"domain_scores_gemma":[0.997557,0.001243363,0.0004807619,0.0001227412,0.0005380874,0.00005793272],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0003384083,0.00003838326,0.0123482,0.009599109,0.0001382594,0.002207634,0.04358352,0.00117714,0.02754069,0.2022799,0.003105243,0.6976436],"study_design_scores_gemma":[0.00004434253,0.0006412765,0.08934435,0.013507,0.0004759217,0.01516867,0.08151165,0.002133458,0.04544514,0.1675897,0.5837211,0.0004174605],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4368103,0.433381,0.04197446,0.002801523,0.0004825098,0.00006553352,0.0004290218,0.0001316347,0.08392411],"genre_scores_gemma":[0.9304928,0.06033514,0.005711167,0.0002829396,0.00008020245,0.00003542478,0.000120596,0.0000276294,0.002914068],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.00362506,"threshold_uncertainty_score":0.01061273,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4294663728","doi":"10.1515/lingvan-2021-0109","title":"Exploring variation and change in a small-scale Indigenous society: the case of (s) in Pirahã","year":2022,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Variation (astronomy); Sound change; Indigenous; Context (archaeology); Scale (ratio); Language change; Social change; Psychology; Demography; Geography; Social psychology; Sociology; Linguistics; Political science; Cartography; Ecology; Biology","authors":[{"name":"Emily Sadlier-Brown","is_ca":true},{"name":"Raiane Salles","is_ca":true},{"name":"Isabel Salomon","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1217047399356564,"gpt":0.3238299199749468,"spread":0.2021251800392904,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002032598,0.0002156264,0.0004015779,0.002447945,0.004209717,0.001475761,0.0008276279,0.0004316739,0.002150697],"category_scores_gemma":[0.006118364,0.00023795,0.0001948014,0.002925252,0.005028941,0.001083634,0.003187366,0.0005683351,0.0001690246],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001241735,"about_ca_system_score_gemma":0.000714482,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02814753,"about_ca_topic_score_gemma":0.06788137,"domain_scores_codex":[0.9979962,0.0008929958,0.00008947121,0.0003390952,0.0003252193,0.0003570303],"domain_scores_gemma":[0.9973006,0.001410186,0.0004497357,0.0003024576,0.0003757388,0.0001614264],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.0001015801,0.00007334479,0.139019,0.0001289913,0.0000374878,0.00281532,0.8142521,0.00009355658,0.007339367,0.003109744,0.000284794,0.03274474],"study_design_scores_gemma":[0.00000449937,0.0002063733,0.490076,0.00004012105,0.00002593121,0.001520723,0.499166,0.0002943856,0.0009886371,0.0009063825,0.00673859,0.0000323092],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.997847,0.00003102869,0.0001682801,0.00006645692,0.000002584525,0.000008405393,0.00001638741,0.000002200989,0.001857706],"genre_scores_gemma":[0.9994436,0.0000199696,0.0002562413,0.00001223794,0.000003004755,0.000007543974,0.00001064689,0.000002457469,0.0002443906],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02814753,"threshold_uncertainty_score":0.05596739,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4402422143","doi":"10.1515/lingvan-2023-0102","title":"Bibliographic bias and information-density sampling","year":2024,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"Marcus och Amalia Wallenbergs minnesfond","keywords":"Sampling bias; Sampling (signal processing); Statistics; Information retrieval; Computer science; Geography; Mathematics; Sample size determination; Telecommunications","authors":[{"name":"Maja Robbers","is_ca":true},{"name":"Harald Hammarström","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02514485621604285,"gpt":0.2974952076023429,"spread":0.2723503513863001,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.2357713,0.0006860632,0.002780643,0.02451419,0.004284187,0.009370446,0.004950193,0.002544479,0.006182659],"category_scores_gemma":[0.6992773,0.00117837,0.001006363,0.04074041,0.01157787,0.01206647,0.007245197,0.002208245,0.001210379],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005876318,"about_ca_system_score_gemma":0.003709164,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006922584,"about_ca_topic_score_gemma":0.005930678,"domain_scores_codex":[0.6123956,0.2940697,0.0239575,0.01956209,0.04731511,0.002699922],"domain_scores_gemma":[0.1295011,0.7637253,0.03291723,0.04990686,0.02281991,0.001129556],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006081325,0.0001675689,0.1112997,0.004715011,0.001001008,0.001227365,0.02224816,0.007240648,0.001059244,0.5850279,0.01309213,0.2523131],"study_design_scores_gemma":[0.0002543362,0.0001829965,0.04888704,0.00485725,0.0005624277,0.002268831,0.01222187,0.02936131,0.00294821,0.8314807,0.06675147,0.0002234393],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2159551,0.01712636,0.6698187,0.02017785,0.001079099,0.00262104,0.0038725,0.0009101343,0.06843914],"genre_scores_gemma":[0.8796324,0.002908094,0.1076712,0.002133375,0.0008125696,0.003153472,0.001341766,0.000149201,0.002197946],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9754858,"threshold_uncertainty_score":0.9424301,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3092479361","doi":"10.1515/lingvan-2018-0068","title":"<i>I feel like</i> and <i>it feels like</i>: Two paths to the emergence of epistemic markers","year":2020,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Toronto","funders":"","keywords":"Complementizer; Preference; Variation (astronomy); Collocation (remote sensing); Vernacular; Persistence (discontinuity); Value (mathematics); Envelope (radar); Psychology; Epistemology; Linguistics; Philosophy; Computer science; Mathematics; Statistics; Syntax; Physics","authors":[{"name":"Marisa Brook","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02804845608191886,"gpt":0.3099939949485337,"spread":0.2819455388666148,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001538128,0.0002003886,0.0002246324,0.001048683,0.002846458,0.004859407,0.0006164528,0.0009032514,0.006586276],"category_scores_gemma":[0.00587343,0.0003843115,0.0001862858,0.0008424344,0.01324624,0.005226535,0.003078386,0.002346479,0.0004151769],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00288988,"about_ca_system_score_gemma":0.002609939,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04462097,"about_ca_topic_score_gemma":0.06236068,"domain_scores_codex":[0.9987877,0.000448021,0.00003258939,0.000232727,0.0002705441,0.0002284299],"domain_scores_gemma":[0.9971343,0.001173834,0.0003628512,0.0003497544,0.0006729613,0.0003062375],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.000259224,0.0000644175,0.02046216,0.0001265231,0.00001291572,0.0008803523,0.5196401,0.0001543466,0.01729424,0.4071391,0.001311635,0.03265494],"study_design_scores_gemma":[0.00006234214,0.0002324652,0.1951473,0.0004547418,0.00006871305,0.002845757,0.5534746,0.002594996,0.006791978,0.1455505,0.09254639,0.0002302477],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8280145,0.0004591141,0.01580372,0.002330635,0.00005435079,0.00004690723,0.00008554976,0.00007248428,0.1531328],"genre_scores_gemma":[0.9973917,0.00004592534,0.001231997,0.00005189016,0.000003693107,0.000006226484,0.00001607252,0.00002627976,0.001226256],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04462097,"threshold_uncertainty_score":0.08872247,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4408918469","doi":"10.1515/lingvan-2024-0247","title":"Asymmetry in French speech-in-noise perception: the effects of native dialect and cross-dialectal exposure","year":2025,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Phonetics and Phonology Research","field":"Psychology","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Asymmetry; Speech perception; Perception; Linguistics; Noise (video); Psychology; Speech recognition; Acoustics; Audiology; Computer science; Physics; Artificial intelligence; Philosophy; Medicine","authors":[{"name":"Scott Kunkel","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.009096022296645455,"gpt":0.3488086382358057,"spread":0.3397126159391602,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006233681,0.0002626655,0.0001969612,0.0005153168,0.0003134984,0.0007039392,0.0001454795,0.0002841458,0.003428724],"category_scores_gemma":[0.00261611,0.0001044542,0.0001322607,0.0001513655,0.0004280311,0.0002275566,0.0005877611,0.0002692154,0.0001951526],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005951148,"about_ca_system_score_gemma":0.000257154,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02837233,"about_ca_topic_score_gemma":0.04165866,"domain_scores_codex":[0.9995295,0.0001252755,0.00002540666,0.0001259456,0.0001157513,0.00007823198],"domain_scores_gemma":[0.9981421,0.0008866579,0.0002790046,0.0001346928,0.0002948225,0.000262636],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.003744933,0.0004640384,0.4959251,0.0001500585,0.0001480025,0.001084175,0.02166887,0.0002057739,0.4367035,0.0002270087,0.0002655036,0.03941308],"study_design_scores_gemma":[0.000003271636,0.000143025,0.9959533,0.000004080449,0.000007158925,0.0001153807,0.001535284,0.00005628498,0.002006037,0.00001867627,0.0001510205,0.00000655618],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9991314,0.0000454692,0.00009455507,0.00001104025,0.000001563641,0.000004400426,0.00003447564,0.000003091811,0.0006741197],"genre_scores_gemma":[0.9995524,0.00002229664,0.00006825975,0.00001154864,9.699475e-7,0.000005392238,0.00003557547,0.000002407261,0.0003012264],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02837233,"threshold_uncertainty_score":0.05641443,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4391967989","doi":"10.1515/lingvan-2023-0038","title":"Plains Cree Order as alternation","year":2024,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Lexicography and Language Studies","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Alternation (linguistics); Phenomenon; Lexeme; Linguistics; Computer science; Mathematics; Epistemology; Philosophy","authors":[{"name":"Atticus Harrigan","is_ca":true},{"name":"Antti Arppe","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01925015524288478,"gpt":0.2696289642022423,"spread":0.2503788089593575,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001507431,0.0002662711,0.0002976473,0.001068743,0.001311704,0.002739618,0.0005211683,0.0006087486,0.004865023],"category_scores_gemma":[0.003605494,0.0002559291,0.0003945638,0.001096118,0.007818437,0.007703538,0.0023537,0.001670845,0.0006373324],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001111386,"about_ca_system_score_gemma":0.0005448047,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008558249,"about_ca_topic_score_gemma":0.001027895,"domain_scores_codex":[0.9979844,0.000789096,0.0001559936,0.0003454938,0.0005542171,0.0001708556],"domain_scores_gemma":[0.9972536,0.0012409,0.0002849447,0.0006920558,0.0004510437,0.00007740434],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0000199711,0.000007708154,0.000396014,0.00003401753,0.000003252379,0.0001434545,0.004291182,0.0001336178,0.001643053,0.9875392,0.0005327642,0.005255743],"study_design_scores_gemma":[0.00001966943,0.00005007408,0.001942105,0.00006320323,0.00001459816,0.001111851,0.003028994,0.003157949,0.004499755,0.8481939,0.1378848,0.00003314568],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2848885,0.001210253,0.2133282,0.004209432,0.0005122224,0.0001699165,0.000396609,0.0007853646,0.4944995],"genre_scores_gemma":[0.9738178,0.0002363595,0.01370045,0.0002065912,0.00008122053,0.00003920936,0.0001149465,0.0002159872,0.01158731],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004865023,"threshold_uncertainty_score":0.01627511,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4399982581","doi":"10.1515/lingvan-2023-0148","title":"The syntax of African American English borrowings in the Louisiana Creole tense-mood-aspect system","year":2024,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Linguistic Variation and Morphology","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Arts and Humanities Research Council; University of Oxford; Georgetown University; Louisiana State University; McGill University; University of Sussex; University of Pennsylvania","keywords":"Creole language; Linguistics; Syntax; English-based creole languages; History; Mood; American English; Psychology; Philosophy; Language assessment","authors":[{"name":"Oliver Mayeux","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01134392482593893,"gpt":0.2920378295530314,"spread":0.2806939047270925,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003199308,0.0001329707,0.0001420934,0.0007169881,0.000828927,0.001188643,0.0002201802,0.0002196424,0.002373743],"category_scores_gemma":[0.0007736661,0.0001644211,0.00007621599,0.0008334769,0.001129243,0.0007937582,0.000949887,0.0003292588,0.0002093215],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001272723,"about_ca_system_score_gemma":0.0004019091,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01392738,"about_ca_topic_score_gemma":0.03214463,"domain_scores_codex":[0.999841,0.0000618297,0.00001023266,0.00003279548,0.00002677983,0.00002735619],"domain_scores_gemma":[0.9995927,0.0001758458,0.0000871673,0.00004695238,0.00008046757,0.0000167882],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.0006173496,0.0001006545,0.1459936,0.0002548614,0.0001193317,0.003801695,0.2302325,0.0009589844,0.177321,0.310992,0.002492493,0.1271156],"study_design_scores_gemma":[0.00005749733,0.0002183595,0.6843331,0.0002334083,0.0001637056,0.003521685,0.09477788,0.007421935,0.02798687,0.03264773,0.1484336,0.0002042001],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9876623,0.00008830562,0.001401653,0.0001358192,0.000002383204,0.00000884772,0.00008948895,0.00003532361,0.01057591],"genre_scores_gemma":[0.9986522,0.00003401951,0.0005713562,0.00001287202,0.000001428621,0.000004495406,0.00003392529,0.00001536239,0.0006744052],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01392738,"threshold_uncertainty_score":0.02769262,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4415076690","doi":"10.1515/lingvan-2025-0160","title":"Linguistic approaches to fake news research are growing and maturing: commentary on a special issue","year":2025,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"","keywords":"Interpretation (philosophy); Sophistication; Context (archaeology); Fake news; Flourishing; Corpus linguistics; Social media; Natural (archaeology)","authors":[{"name":"Maite Taboada","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1733501797692354,"gpt":0.3979844354813963,"spread":0.2246342557121608,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.03316104,0.00165902,0.002416091,0.005445661,0.01558557,0.01666437,0.006843486,0.05961363,0.005936126],"category_scores_gemma":[0.1283467,0.001242216,0.002344748,0.004935324,0.02797507,0.02044297,0.008513766,0.06769149,0.002917077],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.012849,"about_ca_system_score_gemma":0.01869399,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02242069,"about_ca_topic_score_gemma":0.03197537,"domain_scores_codex":[0.9746574,0.01080655,0.00252014,0.00330987,0.006627479,0.002078561],"domain_scores_gemma":[0.7453697,0.2056235,0.007963033,0.00344913,0.03124124,0.006353409],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001999391,0.00001142611,0.0001011475,0.0003302325,0.00001162734,0.0002765183,0.004030543,0.0000231518,0.00006744118,0.009695109,0.9815013,0.00393154],"study_design_scores_gemma":[0.00001846665,0.00002058656,0.0005097178,0.002431572,0.00003028329,0.0003239367,0.007600461,0.0001083623,0.0001416275,0.00851529,0.9802378,0.00006184469],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.000121274,0.007095202,0.00007447832,0.9288376,0.0629287,0.00000407968,0.000017829,0.000006912919,0.000913948],"genre_scores_gemma":[0.006880221,0.013562,0.0003098611,0.7323713,0.2429608,0.00005312734,0.00003913097,0.0001100947,0.003713443],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.966839,"threshold_uncertainty_score":0.1753744,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4393936388","doi":"10.1515/lingvan-2023-0087","title":"A diachronic consequence of intransitivity: structural underspecification and processing biases in Old French","year":2024,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Categorization, perception, and language","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Underspecification; Linguistics; Verb; Grammar; Representation (politics); Object (grammar); Parsing; Corollary; Optimality theory; Computer science; History; Philosophy; Artificial intelligence; Phonology; Mathematics; Political science","authors":[{"name":"Michelle Troberg","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03199731922509109,"gpt":0.3445328847187323,"spread":0.3125355654936412,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001420676,0.0001784317,0.0001996937,0.0009396408,0.000508228,0.001108558,0.0002239457,0.0003445235,0.001964945],"category_scores_gemma":[0.003290112,0.0001405963,0.0001261779,0.0004235806,0.00196331,0.0009118102,0.0004725464,0.0003890889,0.0001712486],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007004421,"about_ca_system_score_gemma":0.000258371,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006939236,"about_ca_topic_score_gemma":0.008316251,"domain_scores_codex":[0.9994505,0.0001624629,0.00005067876,0.0001359313,0.0001268808,0.00007360107],"domain_scores_gemma":[0.9965811,0.001718969,0.0006391541,0.0004692819,0.0004981396,0.00009338879],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0007939718,0.0001371871,0.4369235,0.0002650993,0.0001206734,0.003867702,0.04843355,0.001121767,0.3462495,0.04829541,0.0009563387,0.1128353],"study_design_scores_gemma":[0.00003404169,0.0002580829,0.909145,0.00004155404,0.00009181673,0.005068978,0.01325534,0.002812813,0.03316001,0.02168844,0.0143359,0.0001080637],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9945368,0.00008204853,0.002072674,0.00008654209,0.000003502388,0.000003065788,0.00004295665,0.00002147844,0.003150764],"genre_scores_gemma":[0.9992855,0.00001684192,0.0003304434,0.00002255848,0.000004262416,0.000001103706,0.00002144889,0.000008622247,0.0003092497],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.006939236,"threshold_uncertainty_score":0.0137977,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4377224620","doi":"10.1515/lingvan-2021-0143","title":"How did COVID-19 impact the use of Japanese complex words with <i>masuku</i> ‘mask’ in 2020?","year":2023,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Language, Metaphor, and Cognition","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"","keywords":"Newspaper; Compounding; Coronavirus disease 2019 (COVID-19); Sentence; Computer science; Advertising; History; Artificial intelligence; Sociology; Media studies; Business","authors":[{"name":"Kiyoko Toratani","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08374675119605893,"gpt":0.3525271364932683,"spread":0.2687803852972094,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001908272,0.000302511,0.0002892138,0.0006986914,0.001913591,0.003435716,0.0003981554,0.0005998399,0.003256347],"category_scores_gemma":[0.0041327,0.0001971179,0.0002846373,0.001034421,0.002951142,0.003309246,0.003120498,0.0009885653,0.0005648069],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00203221,"about_ca_system_score_gemma":0.001397544,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02659189,"about_ca_topic_score_gemma":0.03660933,"domain_scores_codex":[0.9989931,0.000323976,0.00006521235,0.0001880506,0.0001794841,0.0002501932],"domain_scores_gemma":[0.9983134,0.0004095044,0.0005568914,0.00009001977,0.0003765493,0.0002537276],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000631697,0.0001447945,0.4593273,0.0006700696,0.0001015511,0.003559364,0.4299242,0.0003977356,0.01581968,0.01411176,0.004966525,0.07034527],"study_design_scores_gemma":[0.00001469996,0.0002654884,0.4768493,0.0002149762,0.00009151398,0.0006189811,0.4611612,0.0008598529,0.002435343,0.005375911,0.05199186,0.0001209474],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9868172,0.0004716783,0.0004869092,0.0009422567,0.00007349659,0.00001335542,0.0001436559,0.00001108342,0.01104048],"genre_scores_gemma":[0.997875,0.0002445205,0.0003056168,0.000148971,0.00001478435,0.000009246871,0.00009744225,0.00001510048,0.001289341],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02659189,"threshold_uncertainty_score":0.05287421,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4413796994","doi":"10.1515/lingvan-2024-0201","title":"Instance memory models as a general computational framework for exploring language processing: bringing the lexicon to life","year":2025,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Lexicon; Computer science; Cognitive science; Natural language processing; Linguistics; Artificial intelligence; Psychology; Cognitive psychology; Philosophy","authors":[{"name":"Brendan T. Johns","is_ca":true},{"name":"Randall K. Jamieson","is_ca":true},{"name":"Matthew J. C. Crump","is_ca":false},{"name":"Michael N. Jones","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0570391012999994,"gpt":0.3250468555101352,"spread":0.2680077542101358,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001645511,0.0004932549,0.0008785061,0.001849669,0.0008534836,0.004918787,0.002205178,0.001591218,0.004721517],"category_scores_gemma":[0.006127164,0.0005120175,0.001562899,0.001280225,0.004146476,0.01197402,0.002586561,0.002345961,0.0006624325],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001398093,"about_ca_system_score_gemma":0.0008835655,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002466757,"about_ca_topic_score_gemma":0.002633308,"domain_scores_codex":[0.9993446,0.0003605201,0.00003410133,0.000123555,0.00007910669,0.0000580435],"domain_scores_gemma":[0.9973186,0.001815724,0.0001448689,0.0004057989,0.0001343284,0.0001806358],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001397093,0.00001405665,0.0003587876,0.00003653947,0.00002162078,0.00006628758,0.0005392333,0.01507808,0.000335898,0.9766784,0.0006716613,0.006185378],"study_design_scores_gemma":[0.0000056034,0.000007543896,0.00006008318,0.00001149599,0.000006798412,0.00003769531,0.00007973708,0.07329762,0.00007448604,0.9247548,0.001656534,0.00000770458],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04727067,0.001141674,0.9266911,0.005514485,0.00007313515,0.00005391774,0.0002667917,0.000357148,0.01863107],"genre_scores_gemma":[0.788472,0.001198642,0.2037794,0.0006211465,0.0002644733,0.0002493557,0.0003153725,0.0001834425,0.004916127],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004918787,"threshold_uncertainty_score":0.01579505,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4403221822","doi":"10.1515/lingvan-2024-0091","title":"Using constructed languages to introduce and teach linguistics","year":2024,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"EFL/ESL Teaching and Learning","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Linguistics; Applied linguistics; Computer science; Quantitative linguistics; Media linguistics; Programming language; Sociology; Philosophy","authors":[{"name":"Cameron Rousseau Duval","is_ca":true},{"name":"Nynaeve Perkins Booker","is_ca":false},{"name":"Chloe Brotherton","is_ca":false},{"name":"Anthony Michael Diaz","is_ca":false},{"name":"Masoud Jasbi","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03780478594028577,"gpt":0.3182800522976544,"spread":0.2804752663573687,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004419092,0.0006256269,0.0001890898,0.0009321545,0.001102796,0.003897032,0.001230863,0.0007680485,0.011791],"category_scores_gemma":[0.008995685,0.0004285382,0.0002623756,0.0003842377,0.003686723,0.00325034,0.006001948,0.00221309,0.002120536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001525101,"about_ca_system_score_gemma":0.002063525,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007130774,"about_ca_topic_score_gemma":0.002041825,"domain_scores_codex":[0.9961863,0.00280497,0.00009106764,0.0003461773,0.0003729156,0.0001985821],"domain_scores_gemma":[0.9869604,0.009533957,0.0005988583,0.001259877,0.0005943,0.00105255],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004398648,0.003526294,0.01869067,0.0008907654,0.00003230511,0.001622896,0.1794335,0.005986759,0.0687165,0.2096971,0.02599737,0.4849659],"study_design_scores_gemma":[0.0002433546,0.001420595,0.005213249,0.0007818907,0.00004017395,0.001518758,0.0521602,0.02008688,0.06122051,0.08970549,0.767428,0.0001809433],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.5343905,0.0003741073,0.3165304,0.005048902,0.0004650121,0.0005642212,0.0002099813,0.004918296,0.1374985],"genre_scores_gemma":[0.7844381,0.0002551653,0.1772463,0.0006659785,0.00004677004,0.0003189059,0.0002306911,0.0004909575,0.03630723],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.011791,"threshold_uncertainty_score":0.0394448,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4392656786","doi":"10.1515/lingvan-2023-0051","title":"The role of syntax in hashtag popularity","year":2024,"lang":"en","type":"article","venue":"Linguistics Vanguard","topic":"Sentiment Analysis and Opinion Mining","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Popularity; Computer science; Syntax; Natural language processing; Semantics (computer science); CLARITY; Artificial intelligence; Linguistics; Information retrieval; Psychology; Programming language; Social psychology","authors":[{"name":"Ming Feng Wan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.008434858255253173,"gpt":0.2652054880703701,"spread":0.2567706298151169,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005209388,0.0002929706,0.0004324371,0.002090381,0.001348916,0.005713111,0.000570661,0.0007019028,0.005563957],"category_scores_gemma":[0.05684137,0.0005787715,0.0003776464,0.001879442,0.00341001,0.01276158,0.002620632,0.001144081,0.0008906251],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001258704,"about_ca_system_score_gemma":0.0007868107,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001922,"about_ca_topic_score_gemma":0.001577364,"domain_scores_codex":[0.9962642,0.002229258,0.0002600609,0.0004569516,0.0005789967,0.0002105913],"domain_scores_gemma":[0.9358897,0.04505179,0.008424304,0.002944708,0.006134855,0.001554689],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00173226,0.0003404788,0.4803508,0.0005418261,0.0003540981,0.0009996953,0.02701298,0.007989755,0.03247224,0.3152876,0.004322663,0.1285957],"study_design_scores_gemma":[0.0002381976,0.000727519,0.4734927,0.0003148902,0.0004420764,0.001953545,0.01918328,0.08089787,0.0124962,0.3956291,0.01417357,0.0004511671],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9446638,0.0003503519,0.02846741,0.001382305,0.00004988844,0.00005345678,0.0003506465,0.0001966821,0.02448554],"genre_scores_gemma":[0.9976904,0.00006143825,0.001624278,0.00003916703,0.00002182978,0.00001202186,0.00007463452,0.0000461415,0.0004298891],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005713111,"threshold_uncertainty_score":0.02755022,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}