P-demotion database is one of the milestones of the Polonez Bis Project, led by the Principal Investigator Katarzyna Janic from October 1, 2022, to September 30, 2024 at Adam Mickiewicz University in Poland. Three project members collected data through grammar mining: Principal Investigator Katarzyna Janic, Co-Investigator Krzysztof Stroński, and Ph.D. student Mohammad Tavakoli. The data collection took place from 2022 to 2023. The database remains open to further refinement and expansion. Any inconsistencies should be reported to Katarzyna Janic at Katarzyna.Janic@amu.edu.pl.

The crosslinguistic database documents the phenomenon of P-demotion across 55 genealogically diverse languages. P-demotion is observed in verb valency-changing alternations where the P argument (object) of a transitive construction loses its status as a core argument. This change in verbal valency may be encoded by a voice marker. The loss of the core status by the P argument manifests differently across languages. For instance, it may involve changes in P-encoding properties, such as indexing and flagging, or the elimination of this argument from the valency slot, which may also alter its indexing on the verb. The phenomenon of P-demotion thus arises from the operation on verbal valency and results in a syntactically intransitive construction, where the demoted P-argument can be expressed as an oblique, incorporated, or fully eliminated.

Based on the multivariate typology by Bickel (2010, 2011, 2015), which posits that every linguistic phenomenon can be decomposed into crosslinguistically comparable variables, the phenomenon of P demotion was systematically analyzed by breaking it down into formal and functional variables. Formal variables encompassed encoding properties of the core arguments A, P, and S in P-demotion alternations, such as flagging (e.g., case, adposition) and indexation (e.g., verb agreement, cross-referencing), along with the presence of voice marking and variation in P-oblique flagging. In contrast, functional variables were related to various types of P-demotion constructions alone and included properties of the P argument such as referentiality, affectedness, animacy, person, and number. The functional variable of referentiality refers to all types of P-demotion constructions, whereas affectedness, animacy, person, and number were applicable only to P-oblique constructions.

Project members regularly cross-checked coding decisions to maintain consistency and accuracy. An additional strength of the database is its extensive range of structural (formal and functional) variables, each accompanied by specific values and comments from different sources. These comments, cited directly from original sources, offer insights into the rationale for particular coding decisions, enabling readers to consult the cited sources for further context on each variable.

The development of the database was a collaborative effort led by Maciej Danielak and Paweł Przybysz, who worked closely with Katarzyna Janic to (i) design and build the database architecture, (ii) process and integrate data into the system, (iii) select and implement appropriate technologies, and (iv) conduct testing to identify and resolve potential errors prior to public release. The Principal Investigator also established a data verification procedure, under which a project member Wojciech Zeyland reviewed and refined data entries to ensure accuracy and address any inconsistencies.

The interactive database serves a dual purpose. First, it provides a comprehensive platform for exploring the grammatical properties (i.e., structural variables) of P-demotion constructions, thereby yielding valuable insights for comparative and crosslinguistic research. Second, it follows contemporary linguistic standards, including reproducibility, data sharing, and reusability.  The access to the database application requires a computer with internet and a web browser.