What is the consensus on how words are formed across cultures (generally)?

Question

I just did a basic exercise of trying to define a new word. Let's take Tibetan for example, but the language for this question doesn't matter. I tried starting at the base components, like /ka/ ཀ and /na/ ན, to make /kana/ ཀན. First of all, there are various definitions by individual authors on thlib.org, so not sure how that factors in. Second, the meaning of the parts often doesn't make the meaning of the whole. In this case ཀ is "primordial" and ན is "sickness" let's say. But you don't get ཀན "primordial sickness" from here, you get Ghana in this case, or in other cases you might end up having two words like "world" and "basket" become "ox" or something like that.

Basically what I'm pointing out is that there doesn't seem to be much relation other than hand-waviness to the parts of a word to the whole. In English and I'm sure in other languages there are things like "bi-cycle", where "bi" is used correctly to mean two, but in many other cases where those higher-level rules don't come into play yet, the roots and stems and such (the "components") don't derive directly from the parts.

So what I'm wondering is what are the rules generally speaking for how words are created across cultures? That is, what is the way to construct new words? At some level is it just arbitrary? Or is there an actual derivation of some sort built in?

Answer 1

There are over 7,000 answers to the question, at present. You can cut the question up into parts -- what are the grammatical prerequisites, and what are the social factors, i.e. how do you sell others on your new invention. English seems to be open to all sorts of neologisms, just check out Urban Dictionary. We can filter out semantic manipulations of existing words (sick, bad, gear, gay) as probably not being the kind of case you are looking for. The simplest case is "apply the rules of the grammar". In Logoori, the word [ʊtavadéé!kéréé] "the one who hasn't cooked for them" can be created by applying the productive rules of the grammar. Chukchi and Kalaallisut have vast grammatical resources for making up words: English word formation is much more restricted (Vietnamese is even more restricted).

One tack to take that isn't just "apply the normal rules of the language" is to take a word from another language, hence the appearance of kalb for "dog" and bint as a disparaging (usually British) term for females, both taken from Arabic. Sticking to English-etymology words, we have blends like "spork" (spoon + fork), made by taking parts of existing words. You can chop off parts of words (burger ← hamburger). There are "prosodic analogy" processes allowing creation of "chocoholic" from "alchoholic".

Sometimes people just make stuff up out of thin air, for example "splang", "snog", possibly "cruft", though some of them have functional explanations ("frell", owing to restrictions on saying certain other words on TV). A basic rule is (generally) that the word has to be pronounceable by speakers of the language, which is why we don't have clicks in made-up words.

There may be grammatical restrictions on what kinds of words can be invented, when some part of speech is a closed class. In Japanese (and other languages) verb is reputed to be a closed class.