So for context, I am occasionally working on a sort of conlang, and asked this question just recently: How to create words which will be unambiguously parsable in a conlang? In there I run into the problem where I have navofodoroda mean either navo-fodo-roda or na-vofodo-roda, as we defined words/meanings for all those "parts". The problem is, when you write it without all that extra whitespace/metadata (of the dashes), you have navofodoroda and you can't tell which one you are using. So that got me to that question, how do you resolve the problem where I have many words in a lexicon, and then it turns out that I run into this sort of conflict way down the road...

That leads me to this question here. How do natural languages like English or Spanish (or any other) deal with creating words so they don't run into this ambiguity problem?

For example, in English we have the word nineveh but I know that that first part isn't nine for some reason. But even so, this word is unique as far as I can tell (I can't think of many counterexamples), so it's not like nineveh has two distinct meanings. I know English has some words with multiple distinct meanings ("tear", for example), but you can tell based on context what it is. But in what other ways do languages work their magic so they don't run into these sort of conflicts? Do they somehow limit the possibility/scope of possible words, so as to prevent these conflicts? If so, how do they do this? (It seems to me like you'd have to study/know every single word in the lexicon to do it appropriately, which makes it seem extremely hard and brittle). Please show how some natural language or a few natural languages deal with this problem.

The only way I can see solving this problem is to have -- like programming -- special "keywords", which are reserved and you can't use any of the related sound patterns of those words. Then that way you can prevent this ambiguity. But that seems nebulous and hard to pinpoint down to me, I don't even know if it's possible. So in my case, na- is the special prefix, so basically no words other than not can contain at all the affix na. But that seems highly limiting, that would mean na is only used like once in the language? That doesn't seem to be realistic / make sense. I don't know tho, how other languages handle this.

  • 1
    An example in English is ''unlockable'', and indeed many words having the structure ''unXable''. In actual language (as opposed to in writing) stress may disambiguate some (in actual language, but not in writing), and context of course may clarify. But there will be cases where the ambiguity can't be avoided. You could have your prefix ''na-'' trigger some phonological change, maybe movement of stress? Commented Dec 21, 2021 at 21:58
  • 1
    There's an old joke about a foreigner reading in an English paper that somebody had been tried for manslaughter.
    – Colin Fine
    Commented Dec 22, 2021 at 22:54
  • @GastonÜmlaut non-lockable removes the ambiguity.
    – Anixx
    Commented Dec 27, 2021 at 16:53
  • @ColinFine At least man-slaughter and mans-laughter is pronunced differently (yes, I'm a foreigner, but I think that what it was about at least).
    – skyking
    Commented Jan 16 at 9:42

2 Answers 2


Natural languages don't do this, ambiguities happen. An example is the German word Staubecken that can be analysed in two ways: Stau-becken "pond, basin, dam reservoir" and Staub-ecken, plural of Staubecke "dust corner".

When the ambiguity becomes too annoying (i.e., the two reading can occur in the same context) speakers will chose another word to express themselves with less ambiguity.


How do natural languages like English or Spanish (or any other) deal with creating words so they don't run into this ambiguity problem?

As jk said, they don't. Suppose I tell you that a door is "unlockable". Does that mean it's (un-lock)-able, able to be unlocked, or un-(lock-able), not able to be locked? Sometimes it's also ambiguous which morphemes are involved. Is "entrance" from enter+ance, or from en+trance?

Natural language has a lot of redundancy in it. Look at how much you can compress English text, for example, using something as simple as Huffman encoding—English text is written in a very inefficient way that has significant redundancy built in. Ths is why yu cn rd txt wth all the intrnl vwls rmvd. Or why you can hold a conversation in whispers, where voicing information is lost. Or over a noisy channel like radio. This redundancy is a feature, not a bug, and in this case it also means that it's usually pretty clear whether "entrance" means a way to enter, or to put someone into a trance. I'd wager you've never actually run into difficulties telling which meaning of "entrance" was intended.

Now, some languages are more explicit about how noun compounds go together. Many Bantu languages, for example, have gender agreement on the connectives that link nouns to other nouns, and this agreement indicates how they connect:

ma-koki m-a yo m-a mw-ana w-a li-boso
6-right 6-of you 6-of 1-child 1-of 5-front
"Your rights as the firstborn child"

The agreement markers indicate that wa liboso "first" (literally "of the front") attaches to mwana "child", while ma mwana wa liboso "of the firstborn" and ma yo "your" (literally "of you") both attach to makoki "rights".

But these aren't strictly necessary, and many other Bantu languages and dialects get by just fine without them.

  • Given that natural languages don't do it, I am still curious if there is a way.
    – Lance
    Commented Dec 21, 2021 at 22:19
  • @LancePollard There is, such as the Bantu example. (This particular one's from Lingála but a lot of Bantu languages show this same agreement.)
    – Draconis
    Commented Dec 21, 2021 at 22:38
  • Interesting, mind if you explain a little bit more in your description how it works? I'm not following.
    – Lance
    Commented Dec 21, 2021 at 23:05
  • @LancePollard Each connective phrase (ma yo, ma mwana, wa liboso) indicates the gender of the noun it's intended to modify. Ma yo and ma mwana start with "ma", indicating they're modifying a noun of gender 6 (makoki). But wa liboso starts with "wa", indicating it's modifying a noun of gender 1 (mwana). So it's unambiguous which noun each phrase attaches to: it's clearly your rights as the firstborn child, not your first rights as a child, the rights of your first child, or anything like that.
    – Draconis
    Commented Dec 22, 2021 at 0:33
  • 1
    Enter-ance and en-trance are pronounced differently (at least in the way I say them). Enter-ance is pronounced /ˈɛnˌt͡ʃɹəns/ and en-trance /ˌɪnˈt͡ʃɹæns/ (sorry, I don't know how to use the stress symbols). This means obviously, there can't be any ambiguity between the two words. Commented Dec 22, 2021 at 18:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.