If human natural language (general communication languages such as English, Thai, Carib, etc.) is ambiguous / obscure (as shown from the Sorites paradox for example), how come people still understand one another, generally in order / good?

  • 2
    I'll hunt for a neat paper I read long ago on this, but the basic argument was that humans have largely similar experiences of reality. Sep 28, 2021 at 6:07
  • By reading between the lines and being empathic and cooperative. Much of natural language is disambiguated by context, shared cultural background and informal cues, see principle of charity and conversational implicatures. And where that is not enough, specialized dialects are developed to minimize ambiguity, as in mathematics, engineering or sciences.
    – Conifold
    Sep 28, 2021 at 7:34
  • @Conifold I think that mathematics is a language on its own. Sep 28, 2021 at 7:35
  • 1
    Dialect is a language. Mathematicians and non-mathematicians alike use mathematical terms (with meanings different from their colloquial ones) and formulas to supplement texts written in natural languages when they aim at precision. The result is a dialect of the base language, mathematical English, mathematical Mandarin, etc. So do professionals and users in other fields.
    – Conifold
    Sep 28, 2021 at 7:48
  • 1
    I was taught (in computational science) that natural language is necessarily vague - if words are defined too loosely then they convey little meaning but if they are defined too tightly then they can seldom be applied.
    – Frog
    Sep 28, 2021 at 8:20

4 Answers 4


Language is as precise as our needs/goals require, we share similar experiences of the world, we try to take shortcuts when possible, and we have particular biology and capacity for language.

Consider the sentence "Would you take this over there?"

Because we share similar experiences and beliefs, we can anchor you, this, and there in similar manners to the world. Anchoring gives these three indexicals and the whole sentence its desired meaning. By anchoring to the world in similar manners we have agreed upon my location, who you are, what object this refers to, and where over there is.

And in a different setting/context, "Would you take this over there?" may refer to a different you, (maybe now I'm speaking to Alice not Bob), and this and over there may refer to moving a box to the next room instead of taking a note across the hall.

But in all cases presumably my goals were sufficed, or I would have changed my phrasing. There is no inherent limit to this process. I can speak about a dreamworld or down to the bare spacetime facts of particle locations in spacetime, because we share similar experiences of the world.

We can imagine anchoring between two people going astray though. That should be expected. We don't share the exact same experiences of the world or think the same. That just means we need further anchoring to communicate our goals. We rarely have a need to distinguish individual sand grains so we may disagree about when a heap no longer is a heap. Biologically and linguistically it is much much quicker to speak of heaps, piles, volumes, heights, weights, etc to quantize or categorize so we don't come to a consensus about a specific number grains equaling a heap. But at no point is our language incapable of describing the number of grains.


If language were precise, we would need specific words for (say) every apple that ever grew since the beginning of time, because every apple is unique to itself. That would be, shall we say, untenable.

Instead, language is the negotiation of categories. If I say "Hand me the apple", you'll hand me the nearest apple on the assumption that's what I mean. If I say "No, I meant that one", you'll put it back and hand me the one I point to. If you don't understand me when I say "Hand me the apple", I will point and say vague things like "Red, round, shiny...", and you will develop the category 'apple' based on my vague descriptions and your perceptions of the thing I ultimately accept from you.

Understanding is constructed, not handed to us on a silver platter.

  • You didn't define precise in that context; if you say that grouping can't be precise than I don't know why you say that. Sep 30, 2021 at 11:49
  • Language is not necessarily the negation of categories; if someone asks me "where are you from" and I reply "from humanity" there might be negotiation (A human who desires one state for all humans) but there might also be violence based on radically different methods of categorizing humans (a human of a world or a human of a country). Sep 30, 2021 at 11:50
  • If you don't understand me when I say "Hand me the apple", I will point and say vague things like "Red, round, shiny...", and you will develop the category 'apple' based on my vague descriptions and your perceptions of the thing I ultimately accept from you. I have already developed that understanding and I suggest to you to respect humans more, not presenting them as even possibly not understanding what an apple is ------ Sep 30, 2021 at 13:40
  • you could say that this is a way for children to learn about physical objects, at least in part I am willing to accept that, but I don't recognize how that justifies your definition of language. Sep 30, 2021 at 13:41
  • @unmarriedinquirer: By definition, a category is reductive: it organized unique objects by focusing on a small set of common characteristics and ignoring uniquenesses. As such, there will always be cases that share those common characteristics but are not part of the category, and objects that ought to be part of the category but lack one or more of those common characteristics. Categories (like peaches) have fuzzy edges. aNd I'm sorry, but not everyone in the world knows what an apple is (just as many people don't know what nattō is. Perspective, please. Sep 30, 2021 at 16:56

The syntax of a natural language is as logical as can be. It is because natural languages are logical that people can both speak these languages and understand each other.

The vagaries of verbal communication are not a consequence of the illogicality of natural languages. They result instead from the practicalities of verbal communication, essentially the fact that people take shortcuts to save both time and energy.

Natural languages are not less logical than the mathematical language, the latter being in effect a restriction and simplification of the formers. It is because it is a restriction and a simplification that it is easier to check the logicality of mathematical statements.

Mathematicians also limit themselves usually to one or two of all existing mathematical disciplines, which limits the vocabulary, which can only make statements easier to check. You cannot do that in the everyday use of a natural language.

Mathematical statements are also not produced on the hoof, as has to be done in all natural languages, and are instead carefully considered before publication.

Mathematics generally is also logically more straightforward compared to, say, philosophy. The logic of mathematical reasoning is in effect quite simple and the concepts used are precisely defined through axioms. The logic of philosophical statements is probably as simple as those of mathematics, but the vocabulary is usually less well defined. This is not due to any alleged deficiency of natural languages, but entirely to the fact that mathematicians make up the concepts they want to use in their axioms, while philosophers often take pre-existing concepts, concepts that result from any number the conversations, of all speakers of the language, sometimes going back several centuries or even millennia.

A philosophical theory is essentially an analysis of pre-existing concepts that are improperly defined because often extremely abstract, for example free-will, God, justice etc., while mathematical theories are essentially logical reasoning to derive theorems from axioms that are usually well understood because more concrete, such as triangles, numbers etc.


When we say language is ambiguous, we mean that the meaning isn't precise. Nevertheless, some information is still communicated.

In most use-cases, such as everyday conversation, we don't require absolute precision. Although I might say, "yes, I understand you," I don't understand you with absolute precision, I only understand you enough to continue the conversation.

For example, if I say, "I was born in the UK," you might not know:

  • exactly where the UK begins and ends (its borders are fuzzy, and whose definitions do you use?)
  • what to do with the fact that birth itself occurs across an unquantifiable amount of space - precisely where did it in fact take place? (it doesn't occur in an infinitesimal point that can be neatly categorised)

But nevertheless if I say I'm born in the UK, despite not being able to understand the meaning with 100% precision, the information communicated is probably enough for a normal conversation. It's an acceptable amount of ambiguity.

When operating in another domain such as law, mathematics or computer programming, and you're evaluating natural language statements, you would typically have to explicitly define concepts and edge cases, so in those situations the tolerance for ambiguity may be much less.

  • For example, if I say, "I was born in the UK," you might not know exactly what it means to be born in one place as opposed to a very nearby place sorry, I totally misunderstand what you meant to say here, please consider to edit. Sep 29, 2021 at 9:49
  • @unmarriedinquirer Thanks, I'll try to edit it to be clearer. Regarding what I meant, places often don't have rigid definitions of where they begin or end, so it's difficult to determine whether a given location really is in a certain place or not, or is in another place, etc. To be able to verify statements of the sort "X happened in Y place", you need to use criteria for what events count as happening inside as opposed to outside that place, and there's usually no non-arbitrary way to do that. Sep 30, 2021 at 10:17

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .