The Nonverbal Element in Persian Verbal Multiword Expressions: A Corpus Annotation Approach

Abstract

This article presents a linguistic framework for the identification and annotation of Persian (Farsi) Verbal Multiword Expressions (VMWEs), developed in alignment with the standards and methodologies set by the PARSEME Corpus—an international research network focused on the systematic analysis of multiword expressions across languages. The study aims to bridge the gap between universal annotation guidelines and language-specific grammatical features by tailoring the PARSEME framework to the structural and semantic properties of Persian. By extracting the characteristics of Persian VMWEs, particularly their nonverbal elements (preverbs) and their diverse syntactic and morphological patterns, this work contributes to a more refined understanding of Persian verbal idiomaticity and the advancement of natural language processing tasks. The article details the development of annotation guidelines that reflect both cross-linguistic categories and Persian-specific grammatical phenomena and the process of annotating a corpus of 5,617 sentences encompassing a wide range of Persian VMWEs including light verb constructions, verbal idioms, and prefix verbs. The practical applications of these guidelines in natural language processing are discussed, highlighting their potential to enhance machine understanding of complex verbal constructions, improve syntactic parsing accuracy, and support downstream tasks such as machine translation, information extraction, and semantic role labeling.