Abstract
These are the guidelines for the application of SNACS (Semantic Network of Adposition and Case Supersenses; Schneider et al. 2018) to Modern Standard Hindi of Delhi. SNACS is an inventory of 50 supersenses (semantic labels) for labelling the use of adpositions and case markers with respect to both lexical-semantic function and relation to the underlying context. The English guidelines (Schneider et al., 2020) were used as a model for this document. Besides the case system, Hindi has an extremely rich adpositional system built on the oblique genitive, with productive incorporation of loanwords even in present-day Hinglish. This document is aligned with version 2.5 of the English guidelines.
Introduction
In Indo-Aryan languages like Hindi and Urdu, small functional words (adpositions) and case markers carry an immense semantic load. A single marker like ko can indicate a direct object, an indirect object, or even the subject of a "dative" construction (e.g., "To me, it seems...").
Existing Natural Language Processing (NLP) tools often treat these markers as purely syntactic. However, to truly understand the "who did what to whom" of a sentence, we need to distinguish between their various meanings. This paper presents the first comprehensive semantic annotation guidelines for Hindi-Urdu adpositions and case markers using the SNACS framework, providing a standardized method for capturing fine-grained meaning in one of the world's most widely spoken language complexes.
Methodology
The authors adapt the SNACS v2.5 framework to the specific grammar of Hindi-Urdu. The system utilizes a hierarchy of 50 semantic labels divided into three top-level categories: CIRCUMSTANCE (time, place, manner), PARTICIPANT (agent, theme, recipient), and CONFIGURATION (possession, characteristic).
A key feature of SNACS is the use of "construals" ($Function \leadsto SceneRole$). This allows annotators to capture cases where a marker’s inherent grammatical function differs from its role in a specific context. For example, if a "Source" marker is used to indicate an "Agent" in a passive sentence, it is labeled as Source \leadsto Agent. Following linguistic tradition, the guidelines target "Layer II" markers (simple case markers like ne, ko, se, kā) and "Layer III" markers (complex postpositions like ke liye "for" or ke bāre meṃ "about").
Mapping Markers to Meaning
The core of the guidelines involves mapping these universal semantic roles to specific Hindi-Urdu grammatical markers. The following table illustrates how common markers are distributed across various semantic categories.
As shown in the table, markers like se are highly polysemous, covering roles ranging from SOURCE and STARTTIME to CIRCUMSTANCE. Conversely, markers like tak are more specialized, primarily indicating GOAL, EXTENT, or ENDTIME.
Key Linguistic Insights
- Semanticizing the Syntactic: The authors argue that even markers often considered purely syntactic, like the ergative marker ne, have semantic content (typically
AGENTorCAUSER) and should be annotated. - Dative Subjects: Hindi-Urdu is famous for "dative subjects" where the experiencer of a feeling or state is marked with ko. The guidelines provide a clear path for labeling these as
RECIPIENT \leadsto EXPERIENCER. - Fictive Motion: The paper provides rules for "fictive motion," where static descriptions use dynamic language (e.g., "The road goes to the mountain"). These are annotated as
GOAL \leadsto LOCUS. - Productivity of Hinglish: The guidelines explicitly acknowledge the "extremely rich" adpositional system of Hindi, which productively incorporates English loanwords within the oblique genitive framework.
Conclusion and Future Work
The primary result of this work is a robust, 59-page manual that provides mapping for 50 universal semantic roles and detailed syntactic tests to help annotators distinguish between ambiguous uses.
Future work will address current limitations, such as the annotation of "unmarked" arguments (nominative/null case) and the complexities of "Force Dynamics"—the way language expresses physical or psychological pushing or hindering. Additionally, while the current focus is on Modern Standard Hindi of Delhi, future refinements may be needed to account for the divergent choices of postpositions in formal and literary Urdu registers.
Summary updated