Subject realization in Uruangnirin

Pegah Faghiri, Eline VisserLanguage 2025 · DOI

Abstract

This study uses a quantitative approach to investigate subject realization in 659 clauses in Uruangnirin, an endangered Austronesian language of eastern Indonesia. In this language, subjects may be referenced in the clause with an NP, pronoun or zero pronoun, and they are optionally indexed on the verb. We establish that the most common combination is to drop the subject in the clause while indexing it on the verb. We find that verbs are more likely to be inflected if the subject of the clause is a speech act participant (SAP), if it is a zero pronoun or an NP, or if it is a patient. When comparing overt subject expression with zero expression we find that non-SAPs, subjects of transitive clauses, new subjects and subjects with high referential distance are more likely to be expressed as an NP. We also identify variables that play a role in the choice of alternative strategies.

Introduction

In many languages, expressing the subject of a sentence isn't a simple binary choice. Uruangnirin, an endangered Austronesian language spoken by approximately 400 people in eastern Indonesia, presents a complex puzzle for linguists. A speaker can refer to a subject using a noun phrase (NP), a pronoun, or nothing at all (a zero pronoun) in the main clause. Furthermore, they can optionally add a prefix to the verb—a subject index—that points back to that subject. This creates six possible combinations for every clause.

This paper investigates the rules of differential subject realization: the cognitive and grammatical pressures that drive a speaker to choose one strategy over another. By analyzing these choices, the study contributes to our understanding of how human brains manage "referential prominence"—the mental effort required to track participants during a conversation.

Methods and Datasets

The study employs a quantitative, usage-based approach to model real-world speech from the Uruangnirin corpus. This corpus consists of audiovisual materials archived with the Endangered Languages Archive, specifically focusing on 10 free narratives from four different speakers.

Map of the Karas Islands in West Bomberai, Indonesia, where Uruangnirin is spoken.

Researchers annotated 1,060 clauses using the GRAID (Grammatical Relations and Animacy in Discourse) and RefInd (Referent Indexing) schemes. After filtering for irregular verbs and ambiguous constructions, a final dataset of 659 clauses was analyzed using Multifactorial Logistic Regression. This statistical modeling allowed the team to isolate variables like "Referential Distance" (how recently a subject was mentioned) and "Topic Persistence" (how often they appear in subsequent discourse).

Key Insights into Subject Realization

The "Zero + Index" Default

The most common strategy, accounting for 70.4% of cases, is to omit the subject in the clause while marking it on the verb. This identifies Uruangnirin as a "pro-drop" language where the verb carries the primary burden of identification.

Formal Avoidance and Phonetic Similarity

One of the study's most interesting findings is that speakers rarely use a pronoun and a verb prefix together. This is likely due to "formal avoidance." As shown in the table of pronouns and indexes, many forms are phonetically similar (e.g., 1PL.INCL ita vs. ta-). Using both would be redundant and phonetically clunky, so the language naturally favors one or the other.

Uruangnirin pronouns and their corresponding subject indexes.

Accessibility and Semantic Signaling

The data supports Accessibility Theory: "heavier" coding (using a full NP like "the man") is reserved for subjects that are harder for the listener to track, such as those that are new to the story or have high referential distance.

Additionally, the study identified a "Patientive Subject Quirk." While subjects are usually agents (doers), Uruangnirin is significantly more likely to use a verb index when the subject is a "patient" (someone something happens to). In these cases, the indexing acts as a "warning flag" for listeners that the subject is playing an atypical semantic role.

Key Results

  • Subject Form: Non-Speech Act Participants (3rd person), subjects of transitive clauses, and subjects with high referential distance are significantly more likely to be expressed as an NP.
  • Verb Indexing: Verbs are more likely to carry a prefix if the subject is a Speech Act Participant (1st or 2nd person), a zero pronoun, or a patient.
  • Topic Persistence: Subjects mentioned frequently in upcoming discourse are actually less likely to be indexed. Once a topic is firmly established as the "star" of the narrative, the speaker can omit the prefix because the referent is already clear to the listener.

Limitations and Future Work

The study faced constraints due to phonetic ambiguity; third-person pronouns (like ia) were often excluded because they sound too similar to the third-person verb prefix (a-). Furthermore, while 659 clauses provided a solid foundation for modeling, larger datasets would be required to capture rarer linguistic phenomena. Future research may explore how contact with Papuan Malay—which lacks verb indexing—is influencing the potential disappearance of these markers in Uruangnirin.

Summary updated

Abstract

Abstract This study uses a quantitative approach to investigate subject realization in 659 clauses in Uruangnirin, an endangered Austronesian language of eastern Indonesia. In this language, subjects may be referenced in the clause with an NP, pronoun or zero pronoun, and they are optionally indexed on the verb. We establish that the most common combination is to drop the subject in the clause while indexing it on the verb. We find that verbs are more likely to be inflected if the subject of the clause is a speech act participant (SAP), if it is a zero pronoun or an NP, or if it is a patient. When comparing overt subject expression with zero expression we find that non-SAPs, subjects of transitive clauses, new subjects and subjects with high referential distance are more likely to be expressed as an NP. We also identify variables that play a role in the choice of alternative strategies.
BibTeX
@inproceedings{faghiri2025subject,
  author = {Pegah Faghiri and Eline Visser},
  title = {{Subject realization in Uruangnirin}},
  booktitle = {Language},
  year = {2025},
  doi = {10.1515/stuf-2025-2025},
}
Figure