Religiolects—language varieties shaped by re- ligious community identity—are low-resource domains often overlooked within high-resource languages. We present the Indo-Religiolect Corpus, the first large-scale dataset for In- donesian religious language variation, con- taining 3 million sentences from over 100 institutional websites representing Muslim, Catholic, and Protestant communities. Fine- tuning IndoBERT demonstrates these religi- olects are computationally distinguishable: Is- lamic Indonesian exhibits high distinctiveness (91.73%), while Catholic and Protestant vari- eties share substantial lexical overlap yet retain detectable shibboleths (86.41% and 86.64%). Our findings indicate a potential for represen- tation collapse: models trained on majority- normative data may default to secular or Muslim-dominant Indonesian, blurring distinct minority voices. We hypothesize that these gaps plausibly translate into downstream fair- ness risks for applications like content mod- eration and automated hiring. This corpus of- fers a template for documenting sub-national varieties, advancing linguistic equity beyond “National Language” benchmarks toward “No Language Variety Left Behind.”