{"id":"W4406119379","doi":"10.3758/s13428-024-02582-2","title":"Expansion of the SyllabO+ corpus and database: Words, lemmas, and morphology","year":2025,"lang":"en","type":"article","venue":"Behavior Research Methods","topic":"Language Development and Disorders","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Centre de Recherche Industrielle du Québec; Université de Montréal; Université Laval; McGill University","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Morpheme; Transparency (behavior); Computer science; Linguistics; Natural language processing; Agreement; Syllabic verse; Artificial intelligence; Speech recognition","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002412893,0.000954146,0.001170723,0.006460652,0.001368207,0.002354712,0.00193143,0.0008495224,0.04984805],"category_scores_gemma":[0.01012085,0.0007279752,0.0006694149,0.004265864,0.0008533879,0.002497619,0.004047153,0.001976996,0.02932186],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009540024,"about_ca_system_score_gemma":0.003658609,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008580064,"about_ca_topic_score_gemma":0.01391076,"domain_scores_codex":[0.9975285,0.000646011,0.0005937789,0.0005401883,0.0005783288,0.0001133288],"domain_scores_gemma":[0.98954,0.005001667,0.0003266688,0.001905736,0.002649997,0.0005758909],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002033978,0.0005491922,0.00916042,0.005184523,0.0001217424,0.001394575,0.002813182,0.001580982,0.04534506,0.01895098,0.4687769,0.4440885],"study_design_scores_gemma":[0.0002966295,0.0001224695,0.02245492,0.0004916913,0.0001206529,0.001904766,0.00103032,0.006197021,0.01870099,0.005216011,0.9433258,0.0001388715],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"dataset","genre_gemma":"empirical","genre_scores_codex":[0.08145742,0.001507343,0.07869572,0.001533593,0.0009816909,0.001316018,0.7616015,0.01986749,0.05303926],"genre_scores_gemma":[0.09196803,0.0006978204,0.1012401,0.0004913612,0.0002236466,0.002453682,0.7897617,0.005046368,0.008117255],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.04984805,"threshold_uncertainty_score":0.1667584,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1710921430824684,"score_gpt":0.5626949910866405,"score_spread":0.391602848004172,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}