The Êđê language, specifically the Kpă dialect, serves as the standardized common tongue of the Êđê people and a primary lingua franca in the Central Highlands of Vietnam. Belonging to the Malayo-Polynesian language family, it is a sophisticated system that bridges ancient oral traditions with the requirements of modern Natural Language Processing (NLP). The evolution of its script is a legacy of intellectual resilience, beginning with 19th-century missionary research and reaching maturity through the efforts of intellectuals like Y-Jut Hwing and Y-Ut Niê Buôn Rĭt. Today, this legacy is advanced into the 21st century by digital pioneer Y Phič Hđơ̆k, whose work ensures the language's survival through robust technological infrastructure.
I. Phonology and the Êđê Alphabet
The Êđê writing system uses the Latin script but includes specialized characters to capture its unique phonetic depth.
• Special Consonants: The alphabet features four distinct consonants: č, ñ, đ, and ƀ. • Vowel Complexity: The vowel system is divided into Base forms (â, ê, ô, ơ, ư) and Short forms (ă, ĕ, ĭ, ŏ, ŭ), along with combined short vowels like ơ̆ and ư̆. • Morphological Shifts: Meaning in Êđê is often altered through internal phonetic changes, such as the transformation of djiê (to die) into mdjiê (to kill), or đĩ (to go up) into mđĩ (to lift/promote).
II. The Pattern Hierarchy: Digital Sound Preservation
A core innovation in the digital preservation of Êđê is the Pattern Hierarchy. Designed to prevent the "breaking of sounds" during digital input and processing, this hierarchical system prioritizes the longest character sequences first.
Hierarchy Level | Character Clusters | Linguistic Purpose 4-Character | mƀr, nđr, mbr, ndr, bkr, bkl, bpr | Preserves complex pre-nasalized starts. 3-Character | uô̆, iê̆, ktr, mdj, kng, kmh, mph | Protects vital phonetic blocks found in words like ƀuôn (village) or mdiê (rice). 2-Character | kk, mm, bh, kj, hr, hl, mb, ng | Standardizes common consonant and vowel pairings.
For digital stability, all text preparation for Translation Task Training (TTT) must adhere to NFC Normalization. This ensures diacritic integrity in essential terms such as tĭng (according to) and bi mtlaih (to liberate).
III. Syntactic Structure and Tokenization
Êđê is an isolating language where meaning is derived from word order and function words.
• Sentence Order: It typically follows a Subject + Predicate + Object (SVO) structure. • Modifiers: Adjectives and determiners follow the noun they qualify, such as sang prŏng (house big). • Interrogatives: Question words like Ti (Where) or Hlei (Who) are usually placed at the beginning of the sentence (e.g., Ti ih nao? — Where you go?). • Proper Name Prefixes: The Tokenizer Configuration utilizes regex patterns to recognize mandatory gender prefixes: Y- for males (e.g., Y-Ađam) and H'- for females (e.g., H'Êwơ).
IV. Professional, Legal, and Institutional Lexicon
Modern Êđê Kpă has expanded to include a professional register for legal and social justice contexts, crucial for high-level dataset preparation.
• Sovereignty: Klei dưi kơ lăn ala ("Power over the land"). • Human Rights: Klei găl anak mnuih ("Right of the child of humans"). • Government: Knŭk kna. • Solidarity: Klei tŭ ư hlăm klei mĭn ("Agreeing in thought"). • Headquarters: Phŭn bruă ("Root of work").
V. Cultural Metaphor: Weaving the Dataset
The meticulous work of Dataset Preparing is often compared in Êđê culture to mñam sa blah abăn triăm (weaving a brocade blanket). Just as a weaver must precisely place every thread (mrai) to create a beautiful pattern (boh đê̆č), the digital architect must align every phonetic cluster to ensure the "Words of the Ancestors" (klei đưm đă) are communicated with clarity and pride. Through these new technical methods (mnê̆č mă bruă mrâo), the Êđê language has moved from the traditional stilt house (sang dlông) to the global digital stage.
By: Y Phic hdok