Heuristics for identifying Japanese-origin words (和語)

Question

Given a word and its reading, I'm interested in having it be automatically be identified as Japanese-origin (和語, but excluding 和製英語 and 和製漢語) or not. False negatives and positives aren't really a problem for my use case, but naturally they should be minimised if they can.

What would be some heuristics that can be implemented relatively easily in a computer program (say, less than 100 lines per rule (using libraries is fine) and doesn't require large amounts of training data)?

Here are some that I've thought up of. Are they correct? Can they be refined?

• Word does not contain "litte" characters (excluding ヶ) nor long vowels (not so sure about this one)
• Number of syllables (which I suppose is equivalent to mora if the above rule is true) is longer than the number of characters in the word
• Word contains at least one kanji or hiragana (to exclude most of the 外来語)

Related: How to tell if a word is Sino-Japanese or Yamato and http://detail.chiebukuro.yahoo.co.jp/qa/question_detail/q1022933605

Answer 1

First of all, if a word contains both kanji and hiragana, it's very likely to be wago. (E.g. 食べる、美しい、…)

If the word consists of kanji only, like you said counting the number of syllables is a good approach. (Of course, on'yomi are to be one syllable, so long vowels, like リョウ, and endings like ク、ン、ツ、チ pass as one syllable, even if the reading is two morae.)

This basically reduces the problem to distinguishing one-syllable (= one-mora) wago, like 手、目、…, from one-mora on'yomi, like 胃、花、…. If I don't know, I usually guess from the sound it contains and you can probably teach your program to do the same, e.g. カ is likely to be on'yomi. There will be false positives like 蚊 or 木, but these rare exceptions could even be hard-coded.

(Of course, there are words for which the wago–kango–gairaigo distinction is ill-defined, like 手帳 or 胃袋. I'm not sure if you want teach your program to identify these as jūbako yomi, yutō yomi, etc.)

I think you can derive the likelihood from the list of jōyō kanji by making a list of all one-mora (rather one-kana) on'yomi and one-mora kun'yomi.

Answer 2

A few rules of thumb to identify kango 漢語 (and thus wago as well) exist:

1. Virtually all i-adjectives (形容詞) and verbs (except for the -する verbs) are wago. Of course, all function words could only be wago. Thus, kango terms are usually nouns, adverbs or na-adjectives (形容動詞). (I'm not sure about gairaigo, but I don't think that they are often used as na-adjectives: most gairaigo are thus nouns or adverbs).

2. In the Sinitic languages (i.e., Chinese), each character almost always stand for one syllable (and have almost always been so). Consequentially, in most Japanese kango terms, each character is usually pronounced with one syllable (i.e., one vowel/diphthong per kanji), which may be one or more morae. Thus, one could tell that 花見｛はなみ｝ is wago, while 花粉｛かふん｝ is kango.

3. Prominent exception to rule 2: in Middle Chinese (from which kango was borrowed), the pronunciation of some characters had the consonant endials -p, -t, or -k. In modern Japanese, many of those characters are pronounced with two syllables, with the second syllable being ツ (from -t) or ク (from -k). (Most of the -p endials, however, have disappeared in modern Japanese). Examples include words such as 国立｛こくりつ｝ (kango). Note that there is a city in Tokyo Metropolis named 国立｛くにたち｝: this name is wago.

4. Not-so-prominent exception to rule 2: in some early Chinese loanwords, rule 2 is violated. Examples include words such as 儀式｛ぎしき｝ (kango), where 式 is read シキ, which is its 呉音 go-on reading. The reason this happens is that many go-on readings are borrowed indirectly through Korean readings of Chinese characters - and also simply because that go-on are ultimately from an early and non-standard dialect of Chinese.

5. Troublesome situation 1: 当｛あ｝て字｛じ｝ , i.e. using kanji solely for their pronunciation (without regard to their logographic meaning). This happens in Japanese from time to time, and sometimes the on'yomi of the characters are used for this purpose, making things really tricky. One tricky example is 沢山｛たくさん｝, of which the etymology is actually unclear (but presumably wago). The kanji here are ateji, and obviously the word has nothing to do with swamps (沢) or mountains (山). However, it looks like a kango word so much.

6. Troublesome situation 2: sometimes kango terms contain complex (sometimes hyōgai 表外) kanji and are often written (partially or completely) in kana. Examples include 綺麗｛きれい｝ and 皮膚｛ひふ｝ (both kango), which are usually written as きれい and 皮ふ respectively, simply because the kanji are difficult and hard to get right.

以上です。