<![CDATA[This article investigates the semantic evolution of reformasi in Indonesian political discourse from 1998 to 2025. Originally a rallying cry for democratic transformation following the fall of Suharto’s New Order, reformasi has increasingly been used to legitimize bureaucratic and technocratic governance. The study applies diachronic embedding analysis to two corpora: legislative transcripts (DPR RI) and Indonesian news articles, spanning five political eras. Preprocessing included lemmatization, tokenization, and multi word expression normalization. Static word embeddings (SGNS) were aligned using Orthogonal Procrustes, and contextual embeddings from IndoBERT were clustered and compared across time slices. Semantic drift was measured through cosine displacement, Jaccard similarity of nearest neighbors, and Jensen–Shannon divergence of contextual clusters. Significant semantic shifts were identified around key political events such as the 2004 direct presidential election, the 2017 UU Ormas, and the 2020 Omnibus Law. Findings reveal that reformasi has drifted from a term associated with democratic rupture to one embedded in administrative discourse. In legislative corpora, it now co occurs with terms like birokrasi, efisiensi, and ASN, while public discourse maintains traces of its original ideological charge. This shift indicates the broader repurposing of political language in post authoritarian governance. The study contributes to computational linguistics and political discourse analysis by demonstrating how embedding based methods can uncover ideological realignments encoded in language. It underscores the value of semantic tracking in transitional democracies and offers a replicable framework for analyzing discursive transformations.]]>
Copyrights © 2025
