Fake-tanuki soup or Fake tanuki-soup?
連濁(れんだく; en: rendaku)is a phonological rule in japanese that makes the first voiceless consonant of a word change into a voiced consonant when used in a compound word. For example, おり + かみ → おりがみ (ori + kami → origami) ("fold" + "paper" → "paperfolding") - the /k/ sound in かみ becomes a /g/ sound (which is the voiced version of a /k/ sound) by adding a voicing mark -> が.
What’s interesting about 連濁 is that native speakers can use it subconsciously as a sort of “order of operations” system for unfamiliar words, like PEMDAS or BIDMAS in maths. A classic example of this is the にせたぬきじる problem[1]. Native speakers can immediately and with confidence tell the difference in meaning between two compound words they have never heard before, despite the only difference being the voicing of a single consonant. Take the three words 偽 (にせ, meaning “fake” or “imitation”), たぬき (tanuki, the Japanese racoon dog), and 汁 (しる, meaning “soup” or “broth”). They can be combined into the following compound words: にせたぬきじる and にせだぬきじる (note the voicing mark, or dakuten, on the latter). Keep in mind, these two words do not exist in ordinary japanese - they’ve been created as part of a linguistics experiment.
You might think the meaning would be ambiguous in those compound words: is it (imitation tanuki)+soup or imitation+(tanuki soup)? Let’s imagine we’re referring to the former. First, we combine にせ+たぬき. There’s a rule that rendaku can’t occur if there’s already a voicing mark in the second component of the compound, but we’re safe here - たぬき has no voicing mark. Therefore, it becomes にせだぬき. Then, we combine にせだぬき+しる. Again, しる has no voicing mark in it, so we’re safe to add it in, and we get にせだぬきじる.
Conversely, let’s say we were referring to fake “tanuki-soup”. First we combine たぬき+しる. This combines safely to たぬきじる. Then we combine にせ+たぬきじる. But wait, the second component does already have a voicing mark, on じ! So we can’t add one to た. Therefore we end up with にせたぬきじる.
That’s a lot of thinking and linguistic hoops to jump through to make up 2 words, but here’s the thing: Japanese native speakers who have never heard these words before can instinctively deduce the difference in meaning with startling accuracy. They correctly determine the meaning of にせだぬきじる as “a broth made from imitation tanuki” and にせたぬきじる as “a fake version of a dish called ‘tanuki soup’”. Even more surprising is the research findings of Shigeto Kawahara, which show that children as young as 9 years old can consistently deduce the difference as well[2]. I think this shows how incredibly powerful the subconscious mind is at learning linguistic rules, and how bad the conscious mind is at learning them!