One Name, Two Sound Systems

01 / English

One name, two readings

Read 御坂美琴 (みさか みこと) aloud as Misaka Mikoto, then as Yùbǎn Měiqín. The characters have not changed, but the name has acquired another rhythm. Japanese gives it two groups of three morae. Mandarin gives it four syllables, two nasal endings, and a sequence of tones that changes when spoken together. What looked like one name on the page has become two quite different events in the mouth.

Personal names offer a small, familiar place to hear a language’s sound patterns. Names from anime, comics, games, and related media provide useful examples. Many readers already know both forms, yet may never have asked why Hakurei Reimu and Bólì Língmèng feel so different. Familiarity lets us begin with something heard and remembered, then ask which parts of that impression we can describe.

Here, “Chinese” means Standard Mandarin; the Japanese discussion uses contemporary standard Japanese, with Tokyo-type pitch accent as its reference. These are present-day comparisons, not reconstructions of historical pronunciation. The examples help us hear structure; they do not establish what the characters’ creators intended.

Our main examples are 御坂美琴 from A Certain Scientific Railgun, 博麗霊夢 and 霧雨魔理沙 from Touhou Project, and the virtual singers 初音ミク and 鏡音リン. Their Japanese readings are given on the respective official pages. (御坂美琴, 博麗霊夢, 霧雨魔理沙, Piapro characters)

Two different operations are present. 御坂美琴 can be read through the pronunciations associated with the same characters in each language. Japanese 霊 is represented here by Traditional Chinese 靈. 初音ミク additionally involves a different written form: ミク is katakana, while 未來 is the Chinese rendering used on Crypton’s Traditional Chinese character page, which also supplies 鏡音鈴. We should not silently turn the Japanese name into an all-kanji spelling and then treat that spelling as the original. The pinyin and structural analyses in this essay are supplied here.

Nor can a character count determine the Japanese rhythm. 琴 is one Mandarin syllable, qín, but contributes two Japanese morae, こと, to Mikoto. 雨 contributes さめ to Kirisame, not an independently chosen あめ pasted after きり. Establish the reading of the whole name before counting its sounds.

01 / 中文

同一個名字，兩種讀法

把御坂美琴（みさか みこと） 先讀成 Misaka Mikoto，再讀成 Yùbǎn Měiqín。字沒有變，名字的節奏卻換了。 日語中，姓與名各佔三拍；普通話中，它有四個音節、兩個鼻音韻尾，以及一組在連讀時 會發生變化的聲調。紙上看似相同的名字，到了口中，便成了兩件頗不相同的事。

人名雖短，卻是聽見一種語言聲音結構的熟悉入口。動畫、漫畫、遊戲及相關作品中的 名字，正好提供了便於比較的例子。許多讀者早已熟悉 兩種讀法，卻未必問過：為何 Hakurei Reimu 和「博麗靈夢」聽起來如此不同？ 從已經聽過、記得的名字出發，我們可以逐一辨認：這份印象中，有哪些結構能說清楚？

本文的漢語比較限於現代標準普通話；日語以當代標準語為主，談到音高重音時， 以東京式重音為參照。這是對現代讀音的共時比較，不是古音重建。例子用來幫助我們 聽見結構，並不據此推斷創作者當初的命名意圖。

主要例子是《科學超電磁砲》的御坂美琴、《東方 Project》的博麗霊夢與霧雨魔理沙， 以及虛擬歌手初音ミク、鏡音リン。各官方角色頁列出了相應的日語讀法。 （御坂美琴、博麗霊夢、霧雨魔理沙、 Piapro 角色介紹）

這裡有兩種不同的對應。「御坂美琴」是在兩種語言中讀相同的漢字；日文「霊」在 本文中文形式中寫作「靈」。「初音ミク」還涉及寫法的改變：ミク是片假名，「未來」 則是 Crypton 繁體中文角色頁採用的中文名稱；「鏡音鈴」也取自該頁。 不能先悄悄把日文名字改寫成全漢字，再把這個寫法當成原名。表中的拼音與下文的 結構分析由本文補出。

漢字數量也不能決定日語的節奏。「琴」在普通話中是一個音節 qín，在 Mikoto 中卻對應兩拍「こと」；Kirisame 的「雨」對應「さめ」，不能把單獨選出的 「あめ」直接接在「きり」後面。必須先確定整個名字的讀法，才能計算它的聲音單位。

02 / English

Three meanings of “open”

A name may sound open, rounded, narrow, or resonant. These impressions become more useful when we distinguish three different properties of its sounds.

四呼: the beginning of the final

Traditional Mandarin analysis describes a syllable through its initial, final, and tone. The final may contain a medial, a main vowel, and an ending. The 四呼 classification concerns its medial, or its main vowel when there is no medial. (Wikipedia: 四呼)

Pinyin letters need interpreting. The u in yǔ represents ü; the final in wèi belongs with uei. Conversely, the i in shi or zhi is not the ordinary vowel of mī. The finals of zhi, chi, shi, ri are conventionally classified as 開口呼. The label therefore does not mean “a sound made with the mouth wide open,” much less “a name that feels expansive.” (Wikipedia: Mandarin finals and their classification)

Nasal endings: how the syllable finishes

An open syllable ends in a vowel; a closed syllable has a consonantal coda. For these Mandarin names, a useful practical distinction is whether a final ends in the nasal consonants -n or -ng. Finals without these endings include both simple vowels and sequences with vowel offglides. (Mandarin finals)

Below, N marks a nasal ending and V marks the absence of one in these examples. V does not mean “one vowel sound”: měi and lái belong in this group too. This notation records a feature of the endings without deciding every theoretical question about glides. The separator marks a name-part boundary, not a required pause.

御坂美琴 repeats a vowel-final–nasal-final pattern. 博麗靈夢 concentrates both nasal endings in its given name. 鏡音鈴 has one on every syllable, whereas 霧雨魔理沙 has none. These are concrete differences that descriptions such as “flowing” or “contained” might respond to.

A nasal ending is not inherently ponderous. A listener may hear its resonance as gentle or continuous. What we can directly record is where the endings occur; their aesthetic effect still needs listening.

Vowel aperture: what happens inside

Vowel height is another dimension. Low a has a more open configuration than high i, u, ü; lip rounding is separate again. In a final such as uan, a rounded medial leads toward a low vowel before the nasal ending. (Wikipedia: Mandarin vowels)

霧雨魔理沙 makes the distinction audible. All five syllables lack nasal codas, but wù and yǔ begin the name with high rounded vowels, while shā ends it with a low vowel. Its syllable-ending pattern is uniform; its internal vowel movement is not. Calling the whole name “open” loses precisely the change that may make it interesting.

02 / 中文

「開」有三種意思

名字可以聽起來開闊、圓潤、緊窄，或富於共鳴。把三種不同的語音性質分開之後， 這些印象才能成為比較準確的描述。

四呼：韻母如何開始

普通話的傳統分析以聲母、韻母、聲調描述音節。韻母可以包含韻頭、韻腹與韻尾； 四呼主要看韻頭，沒有韻頭時則看韻腹。 （維基百科：四呼）

拼音字母需要解讀。yǔ 中的 u 表示 ü；wèi 所屬的韻母是 uei。反過來， shi、zhi 中的 i，並不是 mī 中那個普通的 i 元音。zhi、chi、shi、ri 的韻母，傳統上歸入開口呼。因此，「開口呼」並不等於「把嘴張得很大」，更不 等於「名字聽起來開闊」。（維基百科：普通話韻母及其分類）

鼻音韻尾：音節如何結束

開音節以元音結束，閉音節則有輔音韻尾。對本文的普通話名字，一個實用的區分是： 韻母是否以鼻輔音 -n 或 -ng 收尾。不帶這兩種韻尾的韻母，既包括單元音，也包括 帶元音性尾部滑動的複合元音。（普通話韻母的結構）

以下用 N 表示鼻音收尾，V 表示本文例子中不帶鼻音韻尾的情形。V 不代表 「只有一個元音音素」：měi、lái 也歸在這一組。這只是記錄韻尾特徵的簡寫， 不藉此裁定滑音在所有音系理論中的地位。分隔符標出名字兩部分的邊界，不表示 必須停頓。

「御坂美琴」重複了「非鼻音收尾—鼻音收尾」的模式；「博麗靈夢」把兩個鼻音 韻尾集中在名的部分。「鏡音鈴」每個音節都以鼻音結束，「霧雨魔理沙」則完全 沒有鼻音韻尾。「流動」「收束」之類的聽感，可能就在回應這些具體差異。

鼻音韻尾並不天生沉重。有的聽者反而會覺得它的共鳴柔和、連貫。可以直接記錄的是 韻尾出現的位置；它產生什麼審美效果，仍需回到聆聽。

元音開口度：音節內部發生什麼

元音高低是另一個維度。低元音 a 的發音構形比高元音 i、u、ü 更開；嘴唇是否 圓攏又是另一回事。像 uan 這樣的韻母，先從圓唇的韻頭移向低元音，再以鼻音 收尾。（維基百科：普通話元音的發音描述）

「霧雨魔理沙」讓這個區分很容易聽見。五個音節都沒有鼻音韻尾，但開頭的 wù、yǔ 是高而圓唇的元音，結尾 shā 卻落在低元音上。它的韻尾類型整齊一致，內部的 元音運動卻不一致。只用一個「開」字概括整個名字，恰好會漏掉其中有趣的變化。

03 / English

Tone numbers are not a melody

Mandarin’s four full lexical tones are conventionally numbered 1–4: high level, rising, low or dipping, and falling. The digits name categories, not four musical pitches. Third tone often lacks its citation-form final rise in connected speech; before another third tone it normally takes a rising realization within a sandhi domain. (Mandarin tone sandhi)

御坂美琴 is particularly instructive. Its two third tones meet at the surname–given-name boundary: bǎn | měi. When the whole name is grouped into one sandhi domain, the first of that pair takes a rising realization, yielding the familiar shorthand 4–2–3–2. With a sufficiently strong break after the surname, the grouping changes. The pinyin above records lexical tones; it is not a narrow transcription of every utterance.

博麗靈夢 offers a different repetition: 2–4 | 2–4. The same rise–fall categories recur across both name parts, even though the nasal endings occur only in the second part. To my ear, this makes the Mandarin name feel more paired and recurrent than its Japanese counterpart. That is an interpretation of the pattern, not a measurement of its beauty.

初音未來 begins with two high-level tones and then moves through a falling and a rising tone. It has a clear lexical change of contour, yet we still need an utterance to know its actual timing and pitch range. A quiet introduction, a surprised call, and a sung lyric do not deliver the same name in the same way. Tone is part of the name; intonation is part of what a speaker does with it.

03 / 中文

聲調數字不是旋律

普通話的四個完整詞彙聲調通常編為 1–4：高平、上升、低或曲折、下降。數字表示 類別，並非四個音樂音高。上聲在連續語流中常常不帶單念時的後半段上升；在同一 變調域內，兩個上聲相接，前一個通常變為升調。 （普通話的上聲變調）

「御坂美琴」尤其適合說明這一點。兩個上聲恰好在姓與名的交界相遇：bǎn | měi。 若整個名字連成同一個變調域，這一對中的前字便讀成升調，可以簡記為 4–2–3–2。若在姓後有足夠強的停頓或韻律邊界，組合方式就改變了。上表拼音 記的是詞彙聲調，並非每次實際發音的精密轉寫。

「博麗靈夢」提供另一種重複：2–4 | 2–4。姓與名都重現升—降的聲調類別， 但鼻音韻尾只出現在後半。就我的聽感而言，這使普通話的名字比日語讀法更有成對 往復的感覺。這是對結構的詮釋，並不是對美感的測量。

「初音未來」先有兩個高平調，再轉入降調與升調。詞彙層面有清楚的輪廓變化， 但要知道實際時長與音域，仍然需要一段具體發話。輕聲自我介紹、驚訝地呼喊、 在歌詞中唱出，都不會把同一個名字說成完全相同的樣子。聲調屬於名字的讀音； 語調則參與說話者當下如何使用這個名字。

04 / English

What counts as length in Japanese?

The Japanese mora, or 拍, is a unit of phonological quantity. A short vowel with an optional preceding consonant normally counts as one. A long vowel counts as two; the moraic nasal ん and the first part of a geminate consonant, written っ, each contribute a mora. Small ゃ・ゅ・ょ combines with its preceding kana: きゃ is one mora, while きや is two. These distinctions appear in NINJAL’s mora inventory.

Rin is a useful counterexample to counting either vowels or syllables as though they were morae. リン is normally one syllable but two morae: the nasal counts. Miku also has two morae, distributed over two short syllables. Equal mora counts leave them with different internal structures.

Likewise, Hakurei Reimu and Kirisame Marisa are both 4+3. The written れい in the former may be realized with a long e vowel; its two-mora quantity remains. The latter consists of seven short consonant–vowel syllables. A repeated long-vowel sequence and a succession of short syllables can therefore occupy the same number of morae.

Kubozono’s account of mora and syllable explains why Japanese needs both kinds of structure. A mora count describes phonological quantity, not an instruction to pronounce every unit for exactly the same number of milliseconds. Nor does the absence of a full voiced vowel necessarily remove a mora: vowel devoicing is part of Japanese speech, not evidence that the rhythm has simply disappeared.

04 / 中文

日語用什麼計算長短？

日語的 **mora（音拍，亦稱「拍」）**是音系中的計量單位。短元音連同可能出現的 前接輔音，通常算一拍；長元音算兩拍。撥音「ん」和以小「っ」表示的促音部分， 各佔一拍。小「ゃ・ゅ・ょ」則與前面的假名結合：「きゃ」是一拍，「きや」是 兩拍。國立國語研究所的音拍表明列了這些區別。

Rin 很適合用來反駁「數元音就是數拍」或「數音節就是數拍」的直覺。リン 通常 是一個音節，卻有兩拍：鼻音也計數。Miku 同樣有兩拍，但分布在兩個短音節中。 拍數相同，內部結構仍然不同。

同樣地，Hakurei Reimu 和 Kirisame Marisa 都是 4+3。前者寫作「れい」的 部分可以實現為長 e 元音，而兩拍的音系長度仍保留；後者則由七個短的輔音—元音 音節組成。重複的長元音序列與接連的短音節，可以佔據相同的拍數。

窪薗晴夫對音拍與音節的論述說明了為什麼日語同時需要這兩種結構。 數拍是在描述音系長度，並不要求每拍精確佔用相同的毫秒數。沒有充分發聲的元音， 也未必代表少了一拍：元音無聲化本來就是日語語音的一部分，不能據此 認為節奏單位已經消失。

05 / English

Listen for what repeats

Misaka Mikoto begins both name parts with み. Its broad vowel pattern is:

i–a–a | i–o–o

The two groups have matching length, matching initial mi, and two repeated vowels at their ends. The symmetry is audible without assigning any particular pitch accent. Mandarin Yùbǎn Měiqín does not preserve that pattern, although its VN | VN endings create another one.

Kirisame Marisa has the broad vowel sequence:

i–i–a–e | a–i–a

It also repeats ri and sa across its two parts. These echoes are a more specific observation than calling the name merely “musical.” In Mandarin, the same characters create a quite different sequence of vowels and tones. Written continuity preserves neither every sound nor every kind of repetition.

A vowel list is still a reduction. It omits consonants, treats vowels at a broad level, and tells us little about pitch or delivery. Its value is that it lets us point to one pattern at a time, then return to the whole name to hear how the patterns work together.

05 / 中文

聽見名字中的重複

Misaka Mikoto 的姓與名都以「み」開始。粗略記下元音，得到：

i–a–a | i–o–o

兩組長度相同，都以 mi 開頭，末尾又各有一對相同元音。不必先指定某個音高重音， 也能聽見這種對稱。普通話 Yùbǎn Měiqín 並不保留這個模式，卻以 VN | VN 的韻尾組合建立了另一種對應。

Kirisame Marisa 的粗略元音序列是：

i–i–a–e | a–i–a

姓與名之間還重複了 ri 和 sa。指出這些回聲，比籠統地說名字「有音樂性」更具體。 普通話用同一組漢字，產生的卻是另一套元音與聲調序列。文字延續下來，不代表 每個聲音、每種重複也都延續下來。

元音表仍然是一種簡化。它略去了輔音，只粗分元音，也幾乎不告訴我們音高與表達 方式。它的用途，是讓我們一次指出一個模式，再回到完整的名字，聽這些模式如何 一起起作用。

06 / English

Beyond the romanization

Tokyo-type word accent concerns the presence and location of a lexical pitch fall, including a distinction between accented and unaccented words. It is not a Mandarin-style tone independently attached to each character, and it is not English-style stress. Knowing that a name is 4+3 morae does not yet tell us its accent. (Mora, syllable, and accent)

The mora divisions above consequently carry no invented accent marks. To study a character’s performed accent, we would need to identify the speaker and context in actual dialogue. A shout, a song, and a neutral introduction are different evidence. OJAD can help visualize prosody, but its documentation describes the output for arbitrary text as an estimate. An automatically generated contour does not settle how a particular character name is used. (OJAD’s functions)

Romanization also encourages comparisons between letters that represent different sounds. In Japanese Marisa, sa is さ; in Mandarin Mólǐshā, sh marks a retroflex sibilant. Mandarin q in qín marks an aspirated alveolo-palatal affricate, not the sound an English reader might assign to q. These articulatory distinctions are described in Wikipedia’s account of Mandarin consonants.

Retaining the characters therefore need not retain a name’s texture. A useful comparison follows consonants, vowels, duration, and pitch together, while allowing the writing to show us which identity the pronunciations share.

06 / 中文

羅馬字之外的聲音

東京式詞彙重音關心的是音高下降的有無與位置，包括有重音與無重音詞的區分。 它不是給每個漢字各配一個普通話式聲調，也不是英語式的輕重音。知道一個名字是 4+3 拍，還不知道它的重音如何配置。 （音拍、音節與重音）

因此，上面的音拍劃分沒有附上臆定的重音符號。若要分析角色名字在作品中的實際 重音，需要辨明是哪個說話者、在什麼語境下說的。呼喊、歌唱與平常介紹，是不同 性質的材料。OJAD 可以幫助呈現韻律，但其文件明確把任意文本的輸出稱為推定。 自動產生的輪廓，不能單憑自身決定某個角色名在作品中的讀法。 （OJAD 的四種功能）

羅馬字也容易誘使我們把代表不同聲音的字母拿來直接比較。日語 Marisa 的 sa 是「さ」，普通話 Mólǐshā 的 sh 卻表示捲舌的噝音。qín 中的 q 是送氣的 舌面前塞擦音，也不是英語讀者可能依字母 q 猜出的聲音。 （維基百科：普通話聲母的發音描述）

保留漢字，未必保留名字的聲音質地。有效的比較要同時追蹤輔音、元音、時長與 音高，而文字則讓我們知道：這些不同的發音，指向同一個角色。

07 / English

Sound and association

It is tempting to take a familiar heroine’s name, notice its high vowels or short ending, and conclude that those sounds create femininity or cuteness. But the name comes with a voice, a face, a story, and repeated exposure. Our response to the sound is already mixed with our knowledge of the character.

Research can identify naming conventions without reducing them to universal properties of vowels. Barešová, Nakaya, and Matlach’s study of contemporary Japanese given names finds associations between gender categories and features such as name endings; the final vowel alone is an inadequate explanation. A familiar sequence can acquire an association through usage. That does not make it intrinsically feminine, or tell us the identity of every person who bears it. (Gender-specific features in contemporary Japanese names, 2024)

These characters do not constitute a representative sample of Japanese names, and phonology alone cannot explain their popularity. They do let us ask better questions. Does the repeated mi in Misaka Mikoto contribute to the impression? Does the contrast between the nasal endings of 靈夢 and the vowel endings of 魔理沙 matter to a Mandarin listener? Would somebody unfamiliar with the characters describe them the same way? The first two questions concern audible structure; the third begins to separate sound from learned affection.

A practical listening exercise is to compare the same name in ordinary introductions in both languages. Mark its boundaries, Mandarin lexical tones and endings, and Japanese morae. Then listen for what those notations omit: consonants, actual pitch, pace, and delivery. Describe the difference before judging it.

御坂美琴 does not need the same symmetry in both languages to remain recognizable. Misaka Mikoto repeats its opening and its three-mora shape; Yùbǎn Měiqín repeats its ending pattern while negotiating adjacent third tones. One written name can support more than one kind of coherence. Learning to hear those different kinds is where the phonology becomes interesting.

07 / 中文

聲音與聯想

看到一位熟悉的女主角，注意到她名字裡的高元音或短結尾，很容易便推論：正是 這些聲音造就了女性感或可愛感。但這個名字早已帶著聲線、面孔、故事，以及反覆 接觸的記憶。我們對聲音的反應，已經混入了對角色的認識。

研究可以辨認命名慣例，卻不必把它化約為元音的普遍本性。Barešová、Nakaya 與 Matlach 對當代日本名字的研究，發現名字結尾等特徵與性別類別之間存在關聯； 只看最後一個元音，則不足以解釋這些差異。熟悉的聲音序列可以在使用中取得 聯想，卻不因此天生屬於女性，也不能決定每一位使用者的性別身分。 （Gender-specific features in contemporary Japanese names，2024）

本文幾個角色不構成日本名字的代表性樣本，音系也不能獨自解釋他們為何受歡迎。 但它們讓我們能提出更好的問題：Misaka Mikoto 重複的 mi，是否參與了名字的 聽感？「靈夢」的鼻音韻尾與「魔理沙」的元音收尾，對普通話聽者有什麼差別？ 不認識角色的人，會不會作出相同的描述？前兩個問題關乎可聽見的結構，第三個 則開始把聲音與後天的喜愛分開。

一個簡單的聆聽練習，是比較同一個名字在兩種語言的日常介紹中如何出現。先標出 邊界、普通話的詞彙聲調與韻尾、日語的音拍，再聽記號所略去的部分：輔音、實際 音高、語速與表達方式。先描述差異，再作判斷。

「御坂美琴」不需要在兩種語言中保留同一種對稱，仍能是那個熟悉的名字。 Misaka Mikoto 重複開頭與三拍的形狀；Yùbǎn Měiqín 重複韻尾模式，同時 處理相鄰上聲的連讀。一個書寫形式，可以容納不止一種聲音上的整合方式。 學會聽見這些不同的方式，正是名字音系分析有趣的地方。