{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:2VFNWEJEPODUJ5VCV4RGMZJNW4","short_pith_number":"pith:2VFNWEJE","schema_version":"1.0","canonical_sha256":"d54adb11247b8744f6a2af2266652db711d84199910f7fe7632de314b6316e37","source":{"kind":"arxiv","id":"2101.11043","version":1},"attestation_state":"computed","paper":{"title":"Deep Subjecthood: Higher-Order Grammatical Features in Multilingual BERT","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ethan A. Chi, Isabel Papadimitriou, Kyle Mahowald, Richard Futrell","submitted_at":"2021-01-26T19:21:59Z","abstract_excerpt":"We investigate how Multilingual BERT (mBERT) encodes grammar by examining how the high-order grammatical feature of morphosyntactic alignment (how different languages define what counts as a \"subject\") is manifested across the embedding spaces of different languages. To understand if and how morphosyntactic alignment affects contextual embedding spaces, we train classifiers to recover the subjecthood of mBERT embeddings in transitive sentences (which do not contain overt information about morphosyntactic alignment) and then evaluate them zero-shot on intransitive sentences (where subjecthood c"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2101.11043","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-01-26T19:21:59Z","cross_cats_sorted":[],"title_canon_sha256":"ab7edaad2631f53f88758941dd165c0f6dc7501a9e2a76c2a5d1fd3066c096bc","abstract_canon_sha256":"8d2f605740683495c28985dd9728578e2844f64f604e05f7cb5747b3fa844d99"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:10:08.268889Z","signature_b64":"qvRXh2SkvWVXVq17a4GEEGdpU2VDI0zG9qtWvHTTYI7ttzaZccMMOMdliM3SobC0UYsz7uJfnwnckcJL/PQhCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d54adb11247b8744f6a2af2266652db711d84199910f7fe7632de314b6316e37","last_reissued_at":"2026-07-05T02:10:08.268422Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:10:08.268422Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Deep Subjecthood: Higher-Order Grammatical Features in Multilingual BERT","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ethan A. Chi, Isabel Papadimitriou, Kyle Mahowald, Richard Futrell","submitted_at":"2021-01-26T19:21:59Z","abstract_excerpt":"We investigate how Multilingual BERT (mBERT) encodes grammar by examining how the high-order grammatical feature of morphosyntactic alignment (how different languages define what counts as a \"subject\") is manifested across the embedding spaces of different languages. To understand if and how morphosyntactic alignment affects contextual embedding spaces, we train classifiers to recover the subjecthood of mBERT embeddings in transitive sentences (which do not contain overt information about morphosyntactic alignment) and then evaluate them zero-shot on intransitive sentences (where subjecthood c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2101.11043","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2101.11043/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2101.11043","created_at":"2026-07-05T02:10:08.268478+00:00"},{"alias_kind":"arxiv_version","alias_value":"2101.11043v1","created_at":"2026-07-05T02:10:08.268478+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2101.11043","created_at":"2026-07-05T02:10:08.268478+00:00"},{"alias_kind":"pith_short_12","alias_value":"2VFNWEJEPODU","created_at":"2026-07-05T02:10:08.268478+00:00"},{"alias_kind":"pith_short_16","alias_value":"2VFNWEJEPODUJ5VC","created_at":"2026-07-05T02:10:08.268478+00:00"},{"alias_kind":"pith_short_8","alias_value":"2VFNWEJE","created_at":"2026-07-05T02:10:08.268478+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.30815","citing_title":"When transformers learn \"impossible\" languages, what do they learn?","ref_index":195,"is_internal_anchor":false},{"citing_arxiv_id":"2512.03676","citing_title":"Different types of syntactic agreement recruit the same units within large language models","ref_index":47,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4","json":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4.json","graph_json":"https://pith.science/api/pith-number/2VFNWEJEPODUJ5VCV4RGMZJNW4/graph.json","events_json":"https://pith.science/api/pith-number/2VFNWEJEPODUJ5VCV4RGMZJNW4/events.json","paper":"https://pith.science/paper/2VFNWEJE"},"agent_actions":{"view_html":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4","download_json":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4.json","view_paper":"https://pith.science/paper/2VFNWEJE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2101.11043&json=true","fetch_graph":"https://pith.science/api/pith-number/2VFNWEJEPODUJ5VCV4RGMZJNW4/graph.json","fetch_events":"https://pith.science/api/pith-number/2VFNWEJEPODUJ5VCV4RGMZJNW4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4/action/storage_attestation","attest_author":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4/action/author_attestation","sign_citation":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4/action/citation_signature","submit_replication":"https://pith.science/pith/2VFNWEJEPODUJ5VCV4RGMZJNW4/action/replication_record"}},"created_at":"2026-07-05T02:10:08.268478+00:00","updated_at":"2026-07-05T02:10:08.268478+00:00"}