TBMM Parliamentary Speech Corpus with Speaker and Party Linkage, 1950-2023

A corpus of 2,418,313 speaker turns from the Turkish Grand National Assembly (TBMM) covering the period from 4 January 1950 to 23 April 2023. An analytical core subset contains 136,118 high-confidence floor speeches linked to 2,868 unique MPs across 26 political parties. The dataset was created by Erdem Demirtas, combining OCR archives and official transcripts.

Use Cases

• Analyze political party rhetoric and ideology based on linked party and MP metadata.
• Study gender representation in politics based on gender-annotated MP records.
• Track legislative speech patterns over time based on the 73-year temporal coverage.
• Conduct speaker attribution and disambiguation based on the speaker-turn segmentation and linkage layers.

Strengths

• Contains 2,418,313 speaker turns with a high-confidence analytical subset of 136,118 speeches.
• Links 702,503 speaker turns to a Wikipedia-derived roster of 10,740 MP-term records.
• Covers a 73-year period (1950-2023) of competitive multi-party politics in Turkey.
• Includes targeted corrections recovering over 6,000 multi-word party-group speeches.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full dataset is unknown, which may limit suitability assessment.
• Data may reflect temporal and source bias inherent to the parliamentary record and OCR process.

Provenance

Source

Combines the Turkronicles OCR archive (1950-2018) and official TBMM Word-format session transcripts (2018-2023).

Collection Method

Speaker boundaries detected using a ten-pattern unified parser with decade-aware OCR normalization.

Time Range

1950-01-04 to 2023-04-23

Freshness

Last updated 2026-04-24 19:35:24; freshness should be verified.

Geography

Turkey