Linguist Anvita Abbi studied the language family on Great Andaman (just south of Myanmar in the Andaman Sea) and made a fascinating discovery:
Somehow my extensive experience with all five Indian language families was no help. One time I asked Nao Jr. to tell me the word for “blood.” He looked at me as if I were an utter fool and did not reply. When I insisted, he said, “Tell me where it is coming from.” I replied, “From nowhere.” Irritated, he repeated, “Where did you see it?” Now I had to make up something, so I said, “On the finger.” The reply came promptly—“ongtei!”—and then he rattled off several words for blood on different parts of the body. If the blood emerged from the feet or legs, it was otei; internal bleeding was etei; and a clot on the skin was ertei. Something as basic as a noun changed form depending on location.
The grammar I was piecing together was based primarily on Jero, but a look through Portman's and Man's books convinced me that the southern Great Andamanese languages had similar structures. The lexicon consisted of two classes of words: free and bound. The free words were all nouns that referred to the environment and its denizens, such as ra for “pig.” They could occur alone. The bound words were nouns, verbs, adjectives and adverbs that always existed with markers indicating a relation to other objects, events or states. The markers (specifically, a-; er-; ong-; ot-or ut-; e-or i-; ara-; and o-) derived from seven zones of the body and were attached to a root word, usually as a prefix, to describe concepts such as “inside,” “outside,” “upper” and “lower.” For example, the morpheme er-, which qualified most anything having to do with an outer body part, could be stuck to -cho to yield ercho, meaning “head.” A pig's head was thus raercho.
[My] studies established that the 10 original Great Andamanese languages belonged to a single family. Moreover, that family was unique in having a grammatical system based on the human body at every structural level. A handful of other Indigenous languages, such as Papantla Totonac, spoken in Mexico, and Matsés, spoken in Peru and Brazil, also used terms referring to body parts to form words. But these terms had not morphed into abstract symbols, nor did they spread to every other part of speech.
I envy Abbi's ability to do that. I've spent 8 weeks now trying to sort out Czech grammar and have only just started to understand the accusative case. It makes a guy want to stick with German...