Ideophones are a word class that are understudied by linguists and not well understood. I think I can clarify a few things though. From my understanding, ideophones are a word class especially prevalent in African and indigenous American languages that depict a sensory scene. The sounds of the words somehow depict a visual, audio, smell based, touch based, etc. scene in a listeners mind. For example, dabodabo (this is transliterated without the accents) is a word in an African language that speakers of that language, upon hearing the word, can intimately picture a duck waddling and the sound associated with a duck at the same time. The sounds of the word literally evoke a strong scene in the mind that cannot be described simply as "a waddle". (The word itself means duck. In one of Dingemanse's papers on ideophones, he describes a language as having two dialectal terms for a duck: one is along the lines of quakquak (I forget the precise word) and the other is dabodabo; both imitate the movement/noise of a duck.) Linguists aren't sure why this is, but is believed to be a cross modal effect in the brain.
One thing we know is that ideophones are overwhelmingly duplicative--pikapika in Japanese means sparkle, and English speakers might associate the sounds in pikapika as being bright; I definitely do. Dabodabo is duplicative, pikapika is, and nurunuru (a Japanese word for slimy) is duplicative. When I see/hear this word nurunuru, I do get a sense of sliminess. Basically, the sounds in the words depict a scene that doesn't have to be audio based. Even though duplication isn't a factor in strictly all ideophones, I do think its a good litmus test. Onomatopoetic sounds, a subset of ideophones that strictly represent noises, are duplicative. It is natural for English speakers to say "oink oink" or "quack quack". In a similar fashion, gobble gobble is a very duplicative word. I'm not sure if it counts as an ideophone, but it definitely evokes a stronger image in my mind than, say, a word like crouch. I would never say crouch crouch because it just comes across as weird; and the word itself doesn't evoke a strong image of someone crouching. The word gobble does; in addition to this, gobble sounds similar to wobble and waddle, which while they aren't inherently duplicative, can be duplicated. They also make great sensory words in poems, and their connection to gobble makes me think they might be at least partially ideophonic. In this way, duplication might be a strong indicator of being ideophonic.
Addendum: words like neigh neigh cant be duplicated (at least it sounds weird to me personally), but this might be because the word neigh no longer sounds as close to a horses neiiigh as it did before historical sound change occurred. This is just a personal theory, but as it is a less onomatopoetic word, it may be less ideophonic, as seen by the fact it isn't natural to duplicate it.