The study is intended to describe the divergence of multilingual words in
caption of instagram account of @Mimi.peri 2018. This is a qualitative descriptive
research. The data used in this study were 86 words and 4 phrases in the caption
discourse. The source of the data in this study is the caption discourse of instagram
account of @Mimi.peri 2018. The research used three stages: (1) data provision, (2)
data analysis, and (3) presentation of data analysis results. In the process of
providing the data the simulation method with basic techniques of tapping and taking
notes. Distributional method was used in the process of data analysis with Segmenting
immediate constituents technique (BUL) continued by advanced techniques of deletion
and subtitutions. At the presentation stage of data analysis, it was done using informal
method using ordinary words. From the results of the research there are 13 word
formation processes: (1) the nasalization of "active verbs + -in" to form transitive
verbs, (2) phonetic removal, (affixes, synopsis, apocalypse), (3) the use of other terms,
( 4) substituting the phonemes a with e, (5) substituting the pronounced au with o and
ai with e , (6) foreign language (English) encoding, (7) shortening of words, (
abbreviations, accents, acronyms, contractions) (8) phonetic addition, (protesis,
episthesis, paradox), (9) capitalization of lowercase letters, (10) combination of
letters with numbers, (11) combinations of letters, numbers, symbols and
abbreviation, (12) sound change processes, and (13) sources (Indonesian, regional
languages, foreign languages, English).