Two weeks before publication, the Regional Language Secretariat（秘书处）/,sekrɪ'teərɪət/ sent editor Mira Sen the booklet（小册子）/'bʊklɪt/ in question（所涉及的）/ɪn ˈkwestʃən/. The publicly funded lexicography（词典编纂）/,leksɪ'kɒgrəfɪ/ project was meant to record local vernacular（地方话）/və'nækjʊlə/, yet its draft treated much of the community's lexis（词汇）/'leksɪs/ as error. An automated thesaurus（同义词词典）/θɪ'sɔːrəs/ had replaced colloquial（口语化的）/kə'ləʊkwɪəl/ expressions with formal ones, and it had labelled each recent catchword（流行说法）/'kætʃwɜːd/ or buzzword（时髦词）/'bʌzwɜːd/ as timeless tradition. In light of（鉴于）/ɪn laɪt əv/ the complaints, Mira had to correct the dictionary rather than（而不是）/ˈrɑːðə ðæn/ defend a fluent but misleading product.

She first replayed（重放）/riː'pleɪ/ recordings made on a gramophone（留声机）/'græməfəʊn/, a videophone（可视电话）/'vɪdɪə(ʊ)fəʊn/, and newer equipment. Through headphones（耳机）/'hɛdfonz/, she separated a beeper（发出哔声的设备）/'bipɚ/ from the screech（尖锐声）/skriːtʃ/ of a nearby megaphone（扩音器）/'megəfəʊn/. One elderly speaker had a husky（沙哑的）/'hʌskɪ/ baritone（男中音）/'bærɪtəʊn/ and spoke throatily（以喉音）/'θrəutili/; another shifted from tenor（男高音）/'tenə/ to falsetto（假声）/fɔːl'setəʊ; fɒl-/, while a soprano（女高音）/sə'prɑːnəʊ/ produced a brief trill（颤音）/trɪl/. Some vowels sounded nasal（带鼻音的）/'neɪz(ə)l/ or guttural（喉音的）/'gʌt(ə)r(ə)l/. One contributor would stutter（口吃）/'stʌtə/ because a speech impediment（言语障碍）/ɪm'pedɪm(ə)nt/ affected their fluency, whereas another tended to gabble（说得又快又含糊）/'gæb(ə)l/ and splutter（急促而含混地说）/'splʌtə/. Mira told the review group, “We can describe how a diphthong（双元音）/ˈdɪfθɒŋ/ changes or a consonant becomes syllabic（成音节的）/sɪˈlæbɪk/, but we cannot infer social status from a voice.”

The written sources required a different test. A nineteenth-century scribe（抄写员）/skraɪb/ had left cursive（草写的）/'kɜːsɪv/ notes on a scroll（卷轴）/skrəʊl/, and a later cataloguer called its decorative marks hieroglyphics（象形文字）/,haɪərə'ɡlɪfɪks/. A fragmentary（残缺的）/'frægm(ə)nt(ə)rɪ/ epitaph（碑文）/'epɪtɑːf; -tæf/ used the same marks to inscribe（刻写）/ɪn'skraɪb/ a mason's name. Once the team deciphered（辨读）/dɪˈsaɪfə/ them, it found a trade notation（记号）/nəʊ'teɪʃ(ə)n/ rather than a word. The records also showed that speakers used a lax（松元音的）/læks/ vowel to accentuate（突出）/ək'sentʃʊeɪt; -tjʊ-/ contrast and sometimes objected vocally（出声地）/ˈvəʊkəli/ when an outsider copied it. There was no reason to（没有理由做）/nəʊ ˈriːzən tuː/ turn either feature into a judgement about character.

Mira then rebuilt the entries in alphabetic（按字母排列的）/ˌælfəˈbetɪk/ order. The software had treated a local acronym（首字母缩略词）/'ækrənɪm/ as an anagram（字母异序词）/'ænəgræm/, deleted an apostrophe（省字符）/ə'pɒstrəfɪ/ that marked omission, and refused to hyphenate（用连字符连接）/'haɪfəneɪt/ a genuine compound. Its italic（斜体的）/ɪ'tælɪk/ labels also confused a pronunciation note with a definition. The team corrected the typography（排印格式）/taɪ'pɒgrəfɪ/ in accordance with（按照）/ɪn əˈkɔːdəns wɪð/ recorded usage and had to choose between（在两者之间选择）/tʃuːz bɪˈtwiːn/ alternative spellings only when the sources supported a distinction. The dictionary no longer（不再）/nəʊ ˈlɒŋɡə/ presented editorial preference under the name of（以……的名义）/ˈʌndə ðə neɪm əv/ community authority.

At this stage（现阶段）/æt ðɪs steɪdʒ/, the secretarial（秘书工作的）/,sekrə'teərɪəl/ team halted printing. The secretariat completed the revised paperwork（文书材料）/'peɪpəwɜːk/, while a typist（打字员）/'taɪpɪst/ repaired citations and a scanner（扫描仪）/'skænə/ captured every signed correction. Staff attached a photocopy（复印件）/'fəʊtəʊkɒpɪ/ of each approval, used a stapler（订书机）/'steɪplə(r)/ for temporary sets, and placed the final sheets in a binder（活页夹）/'baɪndə/. A paperweight（镇纸）/'peɪpəweɪt/ kept loose proofs in order during the meeting. The publisher corrected its imprint（出版项）/ɪm'prɪnt/ page, and the designer added a bookmark（书签）/'bʊkmɑːk/ directing readers to the audio archive.

On that basis（据此）/ɒn ðæt ˈbeɪsɪs/, the committee approved the new booklet and released each recording beside its entry. Readers could hear the voice behind a disputed form and see why the editors had retained it. At the launch, contributors found their words without medical labels, invented ancestry, or silent normalization. The observable result was precise: the dictionary documented variation without flattening speakers into stereotypes, so the community accepted it as an accountable record.