📄 Arabic Dialects of Lamomea and Pudahoa at Students in Gontor 4 and 7 Southeast Sulawesi

Summary

This sociolinguistic case study examines how students at two Gontor Islamic boarding schools in Konawe Selatan, Southeast Sulawesi, speak Arabic once it has merged with their local and national languages. Arabic is an everyday campus language there, and contact with Indonesian and regional languages has produced two local varieties: Lamomea at Gontor Putri 4 and Pudahoa at Gontor Putra 7. Using a qualitative, descriptive method based on observed conversations, the authors set three goals: describe each dialect, classify the kinds of code variation in speech, and map the situations in which speakers choose these dialects. They find two types of code-switching, one through translation and one between speech levels. They also find code-mixing in which Indonesian phrases are inserted into Arabic utterances. The dialects are used across many domains, from formal and religious settings to family, trade and casual talk. The study matters because it records how a learned foreign language becomes localized and serves a community both as a practical medium and as a marker of identity.

Key points

• Two speech communities are studied: Lamomea Arabic at Gontor Putri 4 and Pudahoa Arabic at Gontor Putra 7, both in Konawe Selatan, Southeast Sulawesi.
• The method is a qualitative descriptive case study, framed by Hymes' ethnography of communication and Ferguson's view of Arabic dialect variation.
• Local discourse particles such as ka and ji are attached to otherwise Arabic sentences, for example a question about where someone is going, followed by ka.
• Code-switching takes two forms: translation-based switching and speech-level switching. Code-mixing mostly appears as whole Indonesian phrases embedded in Arabic.
• Reported domains of use are formal, kinship, religious, transactional, social environment, informal and non-formal. The dialects serve for internal communication and as a symbol of group identity.

Interesting ideas

• I1. Local sentence-final particles (ka, ji) attach to a foreign-language clause without changing its grammar, so they behave like a separable pragmatic layer on top of the sentence. — For Mova (tokens): the tokenizer and tagger could treat clause-final discourse particles and tag questions ("right?", "eh") as a distinct category, so they do not distort the morphology or lemmas of the core clause.
• I2. The paper separates switching by translation (the same content restated in another code) from switching by speech level (a shift in register or politeness). — For Mova (frames): prag could label a repeated or restated utterance as either a content paraphrase or a register shift, which would help coreference and speech-act detection in dialogue.
• I3. Mixing is described at the phrase level: whole Indonesian phrases are slotted into Arabic sentences rather than single scattered words. — For Mova (tree): when converting or parsing mixed or quoted material, an embedded foreign phrase could be attached as one unit (for example via flat or a foreign-span marker) instead of being parsed word by word.
• I4. Language choice is analysed by social domain (religious, kinship, transactional and so on), which links the variety used to the kind of situation. — For Mova (world): the world model could record the domain or setting of a scene as a state, so later inferences about formality or participant roles have an explicit, explainable source.
• I5. Each finding is grounded in short attested exchanges that come with glosses, instead of resting on abstract claims. — For Mova (learn): evaluation sets for pragmatics could likewise pair minimal real dialogue snippets with their labels, and include particle-free variants as negative controls to check that the particle, not the clause, drives the prediction.