In Ichibanboshi / 一番星 , I mentioned that the kanji is made of two parts: ichiban / 一番 and hoshi / 星 . If I only gave you the two separate parts, you may think the word would be pronounced ichibanhoshi / いちばんほし , and for good reason: because it’s intuitive. Unfortunately, it sounds just a little awkward. For this reason, similar in idea to the French liaison, Japanese has a phenomenon called rendaku / 連濁 .
Rendaku is the voicing of the first syllable of the second part of a compound word. Literally, it means “sequential voicing”. What do I mean by voicing a syllable? In the kana sets, a voiced syllable is indicated by placing a dakuten on the unvoiced kana. For example, ta / た becomes da / だ when voiced. I won’t be going over kana today, but here’s a list of unvoiced sounds and their voiced counterpart:
- k → g
- s → z
- sh/ch → j
- t → d
- ts → dz
- h/f/p → b
A good test to see if a sound is voiced is to place two fingers on your throat while making the sound; if you feel vibrations, it’s probably voiced.
To give some examples,
- ichiban / いちばん / 一番 + hoshi / ほし / 星 → ichibanboshi / いちばんぼし / 一番星
- hiki / ひき / 引き + katari / かたり / 語り → hikigatari / ひきがたり / 弾き語り
Rendaku is common, although it is unpredictable sometimes. Even native Japanese speakers have trouble with it in some more obscure examples, like names. Despite its unpredictability, it has been observed to follow a few rules, each of which predicts where rendaku does not occur. One of the biggest rules is Lyman’s Law, which states that rendaku is not applied when there is a consonant in the second part that is made by obstructing the airflow. Wikipedia gives these examples:
- yama / やま / 山 + kaji / かじ / 火事 → yamakaji / やまかじ / 山火事 (the ji / じ is voiced)
- tsuno / ツノ + tokage / トカゲ → tsunotokage / ツノトカゲ (the ge / ゲ is voiced)
There are other rules, but I haven’t quite wrapped my head around all of them yet.
The easy way to figure out rendaku is to play it by ear: if it sounds right, it probably is right.