Do we have a term for priori knowledge in linguistics?

Question

Broadly speaking, these terms have been introduced throughout history to categorize knowledge:

• A priori, rationalism, deductive reasoning => meaning that we gain new knowledge, only by using existing knowledge (without needing to experience).

• A posterior, empiricism, inductive reasoning => meaning that we gain new knowledge, by experiencing the external world.

Please correct me if I'm wrong.

Based on this dichotomy, I have encountered a lingual phenomenon. There are some words, that when you try to understand their meanings from their constituents you end up guessing its meaning correct. In other words:

• You encounter a new word
• You know the meaning of its constituent parts (morphology, etymology, acronym, initialism, ...)
• You try to guess the meaning of the new word, by combining the meanings of its parts
• You actually check the meaning of the new word in a dictionary
• Hooray! They are the same. You could gain knowledge, only through your existing knowledge.

However, there are words opposite to this:

• You encounter a new word
• You know the meaning of its constituent parts (morphology, etymology, acronym, initialism, ...)
• You try to guess the meaning of the new word, by combining the meanings of its parts
• You actually check the meaning of the new word in a dictionary
• Alas! they are not the same. You could not gain new knowledge based on your existing knowledge.

Examples of this group:
Hypocenter (hypo means under, center means center, thus hypocenter is the place under the source of earthquake => check => no, hypocenter is THE center. Then why on Earth do we call it hyper?)
Recombination (OK, something that was previously combined, for some reason got apart, and now it is combined again. Re is a prefix that means again. => check => no, it means to combine for the first time. Why on Earth?)
Bicarbonate (OK, bi is a prefix meaning two, thus whatever it is, it must have two carbon atoms, C2 => check => no, it has no binary anything in it, its formula is HCO3, OMG)
Money laundering (OK, laundering means washing. Maybe money was dirty for some reason, now they have a process to clean it and reuse it in money circulation in society => check => Oh, it means money legitimization)

Do we have a term for this phenomenon? Of words not meaning what their constituent parts mean?

I personally call this phenomenon Lingual Deductivity and categorize words into Deductive and Non-Deductive and create alternatives in my mind for non-deductive words and I call them MDT (More Deductive Term).

Answer 1

I don't understand what you're saying about those examples, but it looks like you're talking about compositional meaning versus non-compositional meaning. To take a really obvious example, the meaning of the word "cats" is trivially deducible (a compositional function) from the meaning of the parts "cat" and "-s", and the meaning of "cat" is not deducible from something else like "cap" or "bat". An expression like "money laundering" is historically a metaphor, where a related non-literal, non-compositional meaning is conventionally assigned to the expression, but it is not completely unrelated to the parts. There are many kinds of and terms for non-literal language, which is generally about combinations of words. There is much less linguistic vocabulary surrounding structure of single words. The word nomenclature could be used, but this isn't a linguistic term even though scientific nomenclature is an possible example of compositional semantics.

Answer 2

I am aware of the terms transparent (for a predictable meaning) and opaque (for a word form unrelated to the meaning). There is some continuum in the notion of transparency/opaqueness allowing for shades of grey between fully transparent and fully opaque.